76.2www.topit.me

RESEARCHARTICLEOpenAccessNighttimeassaults:usinganationalemergencydepartmentmonitoringsystemtopredictoccurrence,targetpreventionandplanservicesMarkABellis1*,NicolaLeckenby1,KarenHughes1,ChrisLuke2,SachaWyke1andZaraQuigg1AbstractBackground:Emergencydepartment(ED)datahavethepotentialtoprovidecriticalintelligenceonwhenviolenceismostlikelytooccurandthecharacteristicsofthosewhosufferthegreatesthealthimpacts.
WeuseanationalexperimentalEDmonitoringsystemtoexaminehowitcouldtargetviolencepreventioninterventionstowardsatriskcommunitiesandoptimiseacuteresponsestocalendar,holidayandothercelebration-relatedchangesinnighttimeassaults.
Methods:Across-sectionalexaminationofnighttimeassaultpresentations(6.
01pmto6.
00am;n=330,172)overathree-yearperiod(31stMarch2008to30thMarch2011)toEnglishEDsanalysingchangesbyweekday,month,holidays,majorsportingevents,anddemographicsofthosepresenting.
Results:Malesareatgreaterriskofassaultpresentation(adjustedoddsratio[AOR]3.
14,95%confidenceintervals[CIs]3.
11-3.
16;P20.
9%)analysesfocusonnumberofpresentations,notindividuals.
DatawereextractedforanalysisinPredictiveAnalyticsSoftware(PASWW)Version18.
AnalysisusedANOVAfordirectcomparisonsbetweendifferentdailyassaultattendances.
Binarylogisticregression(LR)wasusedforcalculationofadjustedoddsofattendancebydemog-raphywithnon-attendeescalculatedbyage,sex,anddeprivationspecificsubtractionofassaultpresentationsfromnationalmatchingpopulationnumbers.
LRwasusedasthedependentvariablewasbinaryandthecat-egoricalindependentvariablesfulfilledthecriteriaforsuchmodelling[34].
Generalisedlinearmodelling(GLM)isarobusttechniqueformodellingcountdata(e.
g.
,herepresentationsperday)overafixedtimeperiod[35]andwasemployedheretoexamineindependentimpactsofcalendardays,holidays,andsportingeventsonnighttimeattendancelevels.
Althoughalargedataset,overthethree-yearperiodsomecalendareventsoccurjustonedayayear(e.
g.
,NewYear'sEve).
Thus,toreflecttherangeofreliability,confidenceintervalsarepresentedforbothbasicdescriptivestatistics(Table2)andmodelledrelationships(Table3).
Deprivationrateratios(DRRs)werecalculatedastheratioofthemostdeprivedquintile(IMD5)tothemostaffluent(IMD1)forgenderspecificratesinsingleyearofagecategoriesuptoage75years.
Withdataconformingtonormality,comparisonsbetweendeprivationrateratiosformalesandfemalesusedapaired(byage)sampleTtest.
TheHESdatasystemisspecificallycompiledinordertobeusedforplanningandresearchpurposes[30].
TheCentreforPublicHealthiscompliantwiththeHESProtocol[36](whichcoversdataaccessandsharingissues)underthetermsandconditionsofwhichitundertakesworkonHESdatarelatingtotheepidemi-ologyofviolenceandalcoholanddisseminatessuchin-formation.
Consequently,furtherethicalapprovalforanalysesonthisexistingdatasystemwasnotrequired.
ResultsOverall,75.
8%ofpresentationsweremalewith48.
7%ofallnighttimeassaultpresentationsfallingonFridayandSaturdaynights.
Ageatpresentationpeakedinteenagersandyoungeradults(under15years,4.
0%;15–24years,45.
9%;25–34years,23.
8%;35–54years,22.
9%;55yearsorover,3.
4%)withthelowestlevelsofnighttimepresen-tationsinthoseunder15orover54years.
WeekendsandholidaysConsistentwithfindingsfromotherEDandcriminaljusticestudies[11,37,38],nighttimeassaultshereshowstrongweeklypeaksinpresentationsonFridayandBellisetal.
BMCPublicHealth2012,12:746Page3of13http://www.
biomedcentral.
com/1471-2458/12/746Saturdaynights(Table2).
Englandhasaseriesofsetone-daypublicholidayseachyear(seeTable1).
Thenighttimesoftheseholidayswerenotassociatedwithanincreaseinassaultpresentations.
However,theireves(nightsbeforetheholiday)showedsignificantincreasesinassaults(Table2).
NewYear'sEveshowedagreaterincreaseinassaultsthanthatassociatedwithotherbankholidays,whiletheincreaseseenonChristmasEvewasconsistentwithincreaseswithbankholidaysgenerally(Table2).
ConsequentlyChristmasEve,butnotNewYear'sEve,wasincorporatedwithotherbankholidaysinfurtheranalyses,andnightsoftheweekwerecategorisedasSunday-ThursdayandFriday-Saturday.
UsingGLMtotesttheindependentsignificanceofcalendareffect,theimpactsofFriday-Saturday,bankholidayeves,andtheevengreaterimpactofNewYear'sEveonincreasingassaultsweremaintained(Table3).
AnnualandmonthlyeffectsTherewasnosignificantimpactofyear(Table2)andconsequentlythiswaseliminatedfromfurtheranalyses.
Bivariateanalysesidentifiedanincreaseinnighttimeas-saultpresentationsoverthesummermonths(Table2).
Endofmontheffects(theimpactofbeingpaidattheendofmonth)werenotapparenteitherwhenexaminingthefirstorlasttwodaysorthefirstorlastweekofmonths,andthereforewereeliminatedfromfurtheranalyses(Table2).
UsingGLM,monthlyassaultsshowedanoverallnadirinJanuarywithariseoverthesummerperiod,peakinginAugust(Table3).
CelebrationswithoutapublicholidayAnumberofothernationalcelebrations(notlinkedtoholidays)wereincludedintheGLManalyses(Table1).
Halloween,GuyFawkesandStPatrick'snightswereallassociatedwithsignificantlyincreasedlevelsofassaultpresentations(Table3).
However,StGeorge'sDayandValentine'sDaynightshadnosignificantimpact.
Figure1showshowholidayevesandsomenon-holidayrelatedcelebrationsincreasedassaultswithincertainmonthsoftheyear,whentheyoccurredon(a)Fridays-Saturdaysor(b)Sundays-Thursdays.
SportingeventsFigures1aand1balsoidentifytherelationshipbetweenkeysportingeventsandincreasedassaults.
Ofthoseexam-ined,thegreatestincreaseinassaultpresentationswasassociatedwithnationalteam(England)matchesinthefootball2010WorldCup;withpresentationsnearlytrip-lingwhenmatchesoccurredonSunday-Thursdayeve-nings(June;Figure1a).
FinalsoftheFootballAssociationCup,theUnionofEuropeanFootballAssociationsCham-pionsLeague,andRugbySixNationsEnglandmatchesshowednosignificantimpact(Table3).
However,theTable1CalendarandsportingeventsincludedinassaultpresentationanalysesEventDetailsCalendarYear12008/09-2010/11MonthofyearJanuarytoDecemberDayofweekSundaytoSaturdayEnglishbankholidaysAnynationalpublicholidayEnglandbankholidayevesDaybeforeapublicholidayNewYear'seveandday31stDecemberand1stJanuaryChristmaseveandday24thand25thDecemberLastandfirstweekofmonthandlastandfirsttwodaysofmonthExaminingpaydayeffectse.
g.
,celebratingreceivingamonthlywageStGeorge'sday23rdAprilStPatrick'sday17thMarchHalloween31stOctoberGuyFawkes(Bonfire)night5thNovemberValentine'sday14thFebruarySportingEventsFootballAssociationCupAnnualfinalsUEFAChampionsLeagueFinalUnionofEuropeanFootballAssociationsChampionsLeaguefinalFootballWorldCup2010IncludesonlyEnglandmatchesRugbySixnationsIncludesonlyEnglandmatchesannuallySummerOlympicGames20088th-24thAugust20081Yearsrunfrom31stMarchto30thMarchtoaccountfortime-shifteddata.
Bellisetal.
BMCPublicHealth2012,12:746Page4of13http://www.
biomedcentral.
com/1471-2458/12/746Table2Variationswithcalendareventinaveragenumbers$ofpereveningassaultpresentationsacrossEnglishemergencydepartmentservicesCalendareventsAllattendancesMalesFemales95%CI95%CI95%CIDaysMeanLowUpSigMeanLowUpSigMeanLowUpSigAlldays1095301.
4292.
1310.
8227.
3219.
7234.
972.
670.
874.
4HolidaysNoholiday1072301.
7292.
1311.
20.
946227.
7219.
9235.
40.
86972.
570.
774.
30.
940Bankholiday18287.
7236.
6338.
8209.
1168.
0250.
277.
066.
887.
2Christmas3264.
0154.
8373.
2189.
795.
8283.
673.
358.
288.
5NewYear3278.
0161.
6394.
4205.
088.
8321.
272.
766.
478.
9HolidayNoholiday1071296.
0287.
0305.
0<0.
001222.
9215.
6230.
2<0.
00171.
669.
973.
4<0.
001EvesBankholiday18447.
1390.
0504.
1340.
9293.
5388.
4102.
292.
2112.
3Christmas3485.
3422.
2548.
5399.
7323.
2476.
083.
354.
5112.
1NewYear31196.
3975.
01417.
7956.
7765.
11148.
2231.
0185.
4276.
6DayiSunday142230.
2221.
8238.
6<0.
001166.
2159.
6172.
9<0.
00162.
960.
765.
2<0.
001Monday156198.
6193.
2204.
0143.
6139.
5147.
754.
252.
356.
0Tuesday157186.
1181.
0191.
3132.
9128.
9136.
852.
550.
854.
2Wednesday155197.
8190.
7204.
8143.
6137.
7149.
553.
451.
655.
2Thursday151227.
2221.
2233.
2170.
1165.
3174.
956.
054.
357.
8Friday154503.
1491.
8514.
3393.
2383.
7402.
6107.
1104.
5109.
6Saturday156523.
1509.
1537.
0406.
0395.
0417.
0114.
3111.
0117.
6CalendarJan92246.
0219.
7272.
3<0.
001184.
9163.
5206.
40.
00359.
954.
964.
9<0.
001MonthiFeb84280.
8249.
0312.
6212.
3186.
6238.
166.
960.
873.
0Mar93267.
1240.
2294.
0200.
8178.
8222.
865.
660.
570.
7Apr86309.
7277.
2342.
2231.
9205.
8258.
175.
569.
181.
8May87334.
0300.
1367.
8251.
3223.
4279.
180.
374.
286.
3Jun90321.
2287.
5354.
9238.
8211.
7266.
080.
574.
087.
0Jul93319.
2287.
8350.
6238.
5213.
1263.
979.
373.
185.
5Aug90322.
7293.
2352.
3241.
2217.
0265.
480.
274.
685.
8Sep90297.
5266.
0329.
1224.
8199.
0250.
771.
465.
677.
3Oct93312.
0276.
4347.
6238.
5209.
5267.
472.
465.
679.
2Nov90274.
0244.
2303.
8205.
8181.
6230.
067.
161.
372.
9Dec83266.
7236.
6296.
7205.
7180.
6230.
859.
654.
464.
9ChangeofNeither935296.
6287.
0306.
10.
205223.
6215.
8231.
30.
18971.
569.
773.
80.
257monthi,iiFirst2days69269.
7237.
9301.
6200.
1174.
5225.
768.
461.
974.
9Last2days67314.
7271.
7357.
8236.
9201.
8272.
076.
468.
484.
5ChangeofNeither599294.
2282.
3306.
10.
288222.
0212.
3231.
70.
29570.
868.
573.
00.
203monthi,iiFirstweek246288.
1269.
9306.
2215.
7201.
2230.
370.
967.
274.
6Lastweek226309.
1288.
2329.
1233.
0216.
0250.
174.
770.
778.
6Yeariii2008/09366285.
5270.
1300.
80.
060215.
0202.
5227.
50.
07869.
866.
972.
80.
0672009/10365309.
1292.
1326.
0233.
1219.
4246.
973.
069.
876.
22010/11364309.
8293.
5326.
2233.
9220.
6247.
274.
971.
878.
1$NationalemergencydepartmentdataarenotcompleteforEnglandandthereforefiguresdonotrepresentnationaltotals-seeMethods.
Statisticsuseanalysisofvariance.
iVariablesexcludebanksholidayandChristmaseves.
iiChangeofmonthvariablesmarkthefirstandlasttwodaysandfirstandlastweekineachmonth.
iiiYearsrun31stMarchto30thMarchtoaccountfortime-shifteddata.
Bellisetal.
BMCPublicHealth2012,12:746Page5of13http://www.
biomedcentral.
com/1471-2458/12/7462008Olympicswereassociatedwithasmallbutsignifi-cantfallinassaultpresentationsandthiseffectwasmain-tainedevenwhenGLManalysiswaslimitedto2008(X2=5.
733;P<0.
05).
Theoverallcalendarmodel(Table3)includingnightoftheweek,holidayeves,celebrations,andsportingeventswasastrongpredictorofobservedvalues(observedvmodelledestimates;R2=0.
918;P<0.
001).
SexanddeprivationeffectsNumbersofassaultpresentationsweresignificantlyhigherformalesthanforfemales(Table2;t=50.
84,P<0.
001).
Maletofemaleassaultsratioswerehighestonpeakdaysforassaults;increasingfromameanof2.
76(95%CIs2.
72-2.
80)onSunday-Thursdaynightsto3.
64(95%CIs3.
59-3.
69)onFriday-Saturdaynights(t=24.
40;P<0.
001),andfurtherto4.
14(95%CIs3.
86-4.
43)onNewYear'sEve.
Examiningassaultsbydeprivationiden-tifiedthattheproportionofassaultpresentationsonSunday-Thursdaynightsincreasedwithdeprivation(mostaffluenttomostdeprivedquintile,46.
02%,47.
21%,49.
49%,51.
41%&54.
23%;Xtrend2=1065.
22,P<0.
001).
LRanalysisexaminedtheindividualcontributionofage,sex,anddeprivationtolikelihoodofpresentingintheEDforanighttimeassault(Figure2).
Riskspeakedstronglyatage18years.
Oddsofassaultpresentationweremorethanthreetimeshigherformales(3.
14,95%CIs3.
11-3.
16;P<0.
001)thanforfemalesandnearlyfourtimesgreaterforthemostdeprivedquintilethanthemostaffluent(3.
87,95%CIs3.
82-3.
92;P<0.
001).
Table3GeneralisedlinearmodelexaminingindependentimpactsofcalendareventsonmeannumbersofassaultpresentationspernighttoemergencydepartmentservicesinEnglandVariableSlope(B)95%CIsPMonthsJanuary31.
2244.
8817.
56<0.
001February13.
640.
6927.
980.
062March6.
147.
7620.
030.
387April41.
0227.
1854.
86<0.
001May52.
4438.
6166.
27<0.
001June46.
9933.
0460.
93<0.
001July51.
0337.
3564.
70<0.
001August59.
8845.
5374.
24<0.
001September36.
5422.
7550.
32<0.
001October31.
1117.
3244.
90<0.
001November6.
437.
4720.
340.
365December(Ref)Weekends&Fri-Sat303.
13296.
84309.
42<0.
001HolidaysBankHolidayEvesSun-Thur242.
74221.
55263.
94<0.
001NewYear'sEveSun-Thur970.
35904.
721035.
98<0.
001BankHolidayEvesFri-Sat275.
85183.
57368.
13<0.
001NewYear'sEveFri-Sat1107.
851015.
571200.
13<0.
001Sun-Thur(Ref)iSportingFACupfinal53.
620.
45107.
680.
052EventsSixNationsRugby7.
6518.
4133.
720.
565WorldCupFootballEngland298.
05251.
10344.
99<0.
001Olympics38.
0762.
7013.
440.
002UEFAChampionsfinal32.
0321.
8685.
920.
244iCelebrationsStPatrick'sday65.
3811.
49119.
260.
017Halloween191.
99138.
10245.
88<0.
001Valentine'sday12.
6168.
6143.
400.
659GuyFawkesnight96.
3742.
50150.
24<0.
001StGeorge'sday7.
4546.
4361.
330.
786iSportingeventsandcelebrationsareallenteredintothegeneralisedlinearmodelasseparatebinaryvariables.
Referencecategorieshavebeenomitted.
Datacoverthetimeperiod31stMarch2008to30thMarch2011(seeMethods).
95%CI=95%confidenceintervals.
Fulldetailsofcalendarevents,sportingeventsandcelebrationsaregiveninTable1.
Bellisetal.
BMCPublicHealth2012,12:746Page6of13http://www.
biomedcentral.
com/1471-2458/12/746Inordertoexaminehowtheimpactofdeprivationvarieswithage,deprivationrateratios(DRRs;rateinmostdeprivedquintile/mostaffluentquintile)werecalculated(Figure3a,b).
DRRsinmalesshowedapre-pubescentpeakaroundagesixyears,anadirat21yearsthenanincreasetoapost-adolescenceplateaufromapproximately45years(Figure3a).
VariationsinDRRsinfemalesweresimilartomales.
However,boththepre-pubescentriseandpost-adolescenceplateauwerelesswelldefinedthanformales(Figure3b).
YearwisepairedcomparisonofDRRsbetweenmalesandfemalesidentifiednosignificantdifference(t=0.
688,P=0.
493).
DiscussionWehavedemonstratedthatanationalEDmonitoringsystemcanusefullyidentifyindividualandcommunityFigure1Meanassaultpresentationspernightbymonthandforholidays,sportingevents,orothercelebrations;a)FridayandSaturdaynights;b)SundaytoThursdaynights.
95%confidenceintervalsarepresentedforeachmonthonly.
ForstatisticalsignificanceofdifferencesseeTable3.
Fulldescriptionsoftheholidays,sportingevents,andcelebrationsincludedaregiveninTable1.
BHE=Bankholidayeve;}NewYear'sEvehasmeanvalueprovidedasitisoutsidetheyaxisscale.
Bellisetal.
BMCPublicHealth2012,12:746Page7of13http://www.
biomedcentral.
com/1471-2458/12/746riskfactorsforassaultandchangesinservicepres-sureswithcalendar,celebration,andsportingevents.
Routineanalysesofassaultdataoftenusespolicerecordedcrimetoexaminetheimpactofcalendareventsonassaults.
However,suchdatacanbecon-foundedbybothlevelsofpoliceactivity(numberofindividualsworkinginanyarea),policingpolicy(e.
g.
,whichviolenteventswarrantwarningandwhicharrests)andwheretheytakeplace(e.
g.
,detectionofassaultsinpublicvs.
privatespaces).
EDdataarenotdirectlyimpactedbysuchconfounders,provideameasureofhealthharmsrelatingtonighttimeassaultsandincludeeventsthatarenotreportedtopolice[11].
Figure2Logisticregressionmodelforriskofnighttimeassaultpresentationbyage,sexandquintileofdeprivation.
Modelusesbackwardconditionallogisticregression.
Age,sexanddeprivationquintilesvariablesallmadehighlysignificantcontributionstothemodel;Waldstatistic=220912.
72,76813.
16,69763.
54respectively;P<0.
001foreachvariable.
Bellisetal.
BMCPublicHealth2012,12:746Page8of13http://www.
biomedcentral.
com/1471-2458/12/746UsingsuchEDdatathisstudyidentifiesthatnightsprecedingwork-freedaysseemorethandoublelevelsofassaultspresentations(Figure1a,b).
AssaultlevelspeakinsummermonthsfallingtoalowinJanuary(Figure1a,b;Table3),whenalcoholconsumptioncanalsoreachitsnadir[39].
Althoughviolencehasbeenlinkedwithwarmerweather[40],aconcentrationofindividuals'per-sonalholidaysinthesummerperiodmayalsobeacon-tributingfactordespitemanyindividualsholidayingabroad[41].
Constraintsthatemploymentplacesonthelengthofnightsoutandalcoholconsumptionareremovednotonlybyholidaysbutalsobyunemployment[42].
Thus,themostdeprivedcommunitiesshowedthehighestassaultratesandagreaterproportionofassaultsonSunday-Thursdaynights;consistentwithmoreindivi-dualshavingnoemploymentpressuresmidweek.
Fur-ther,whiledeprivedandaffluentmalesbothshowedpeaksinassaultsratesintheirlateteens,ratesreducedmorerapidlyinthemostaffluent(Figure3a).
Movementintoemploymentinpost-adolescencecanreduceexces-sivealcoholuse[43]–althoughhowthisimpactsonex-posuretoviolenceislesswellstudied.
Wealsoidentified(A)Males(B)FemalesFigure3Affluentanddeprivedquintileassaultpresentationratesianddeprivationrateratiosbyage.
DRR=deprivationrateratio.
DRRsmoothediscalculatedasafiveyearrollingaverage.
iNotallEDservicescurrentlyreportviolencedatatothenationaldatabaseandthereforethesearepresentedonlyforcomparativepurposes.
Bellisetal.
BMCPublicHealth2012,12:746Page9of13http://www.
biomedcentral.
com/1471-2458/12/746deprivation-relateddifferencesinassaultpresentationatearlyages.
Critically,byage15yearsmalesinthemostdeprivedquintilehadexceededthepeakpresentationlevelachievedinthemostaffluentquintileatage19years(Figure3a).
Worse,forfemalesassaultpresenta-tionsintheaffluentquintilepeakedatage20yearsandratesinthemostdeprivedquintileexceededthispeakbyage13years(Figure3b).
Globally,socialinequality,povertyandyouthun-employmenthavebeenassociatedwithincreasedvio-lence[44]andevenriotinginsomecountries,includingtheUK[45].
Internationally,attentionhasfocusedonbothimmediatepolicingmeasurestopreventfurtherviolenceand,increasingly,theneedforlonger-termmultidisciplinarylifecourseapproachestoimproveyoungpeople'sprospectsandreducetheiroverallpro-pensityforviolence[1,17,46,47].
Ontheformer,somelocalEDdatasystemshavealreadybeenusedtorecordassaultlocationandinformthetargetingofpoliceactiv-ity[11-13]ofteninnightlifeareas.
Onthelatterhowever,potentialrolesforEDdataremainlargelyunderdevel-oped.
Resultspresentedhereidentifyamuchearlieres-calationinviolenceinthepoorestcommunitiesandapeakatamuchhigherlevel(Figure3a,b).
Earlylifeex-posuretoviolencerepresentsadirectrisktochildren'simmediateandlong-termphysicalandmentalhealth[48,49];insomecircumstancesresultinginpermanentdisability.
Moreover,suchexposurealsoleavesindivi-dualsmorelikelytoengageinviolencelaterinadoles-cenceandadulthood[48,50].
Earlylifeexposuretoassaultscanbereducedthroughparentalsupport,pre-schoolenrichment,andsocialdevelopmentprogrammes[16-19].
Severaloftheseprogrammes,suchasNurseFamilyPartnerships[16],havealreadybeguntobescaledupinanumberofcountriesincludingCanada[51],Australia[52]andEngland[53].
HerewehaveidentifiedhowanationalEDdatasystemcanprovideabenchmark;identifyingareasmostinneedofsuchinterventions.
Aswellasaroleintargetedlong-termprevention,wehaveshownthatanationalEDdatasystemidentifiespeaksandtroughsinviolencethatarestronglyasso-ciatedwitheventssuchascelebrationsandsportingevents.
InEnglandHalloween,GuyFawkesNightandStPatrick'sDayarenowheavilycommercialisedeventswiththemedalcoholpromotions,organisedpublicevents(suchasclubnightsandbarcrawls)andprivateparties.
Allthreewereassociatedwithsignificantincreasesinassaults(Figure1a,b;Table3).
However,Valentine'sDayandStGeorge'sDayshowednosignifi-cantincreases.
Sportingeventsalsovariedinimpactonassaultpresentations.
Presentationsincreaseddramatic-allyonnightswhenthenationalteamplayedintheWorldCupbutnotwithotherfootballorrugbyfixtures(Figure1a,b;Table3).
Theassociationbetweensportandpublicviolencehasbeenexaminedelsewhere[54,55].
However,thisstudyidentifieshowEDdatacanmeasuretheimpactsofviolencebeyondthattypicallyobservedaroundcitycentresandgatheringssuchassportingevents.
Thus,broadcastaccesstotheBeijingOlympicswasassociatedwithasmallbutsignificantre-ductioninoverallassaultpresentationsinEngland(Table3).
WhileunderstandingsuchpatternsexposesexpectedpressuresonEDdepartments,theyarealsopertinenttootherfrontlineservicessuchasambulanceandpolice.
Currently,thereislittleinformationonhowwellemergencystaffinglevelsareattunedtodemandandnationalEDdata,withlocalintelligence,couldhelpinformtheefficientdistributionofstaffandotherresourcesonacalendarbasis.
Ouranalysisonlyexaminedpublicholidays,nationalcelebrations,andsomemajornationalandinternationalsportingevents.
Inplanningholidaysandeventsnationallymorethoughtshouldbegiventohowselectionofspecifictimes,days,andmonthscouldbeusedtominimiseanyresultantincreaseinviolence.
Moreover,healthandotheragenciesshouldconsidersuchintelligencewhentimingcampaignstoreducebingedrinkingandrelatedviolence,stipulatinglicenserequirements,andenforcingcriticalle-gislation(forexample,nosalesofalcoholtothoseunder-ageoralreadydrunk).
TheabilityofEDdatatoprovideintelligenceonnight-timeassaultsreliesonindividualsreportingviolenceasthereasonfortheirpresentation.
Suchreportingmaybeaffectedbyissuesofconfidentiality.
Moreworkisrequiredonprotectingconfidentialitybyestablishingoptimallevelsofdataaccessfordifferentorganisationsandatdifferentgeographicallevels[56].
WhilesomelocalEDsystemsinEnglandcollectandshareinformationonassaultlocation,fewshareinformationonresidence;despitethisbeingroutinelycollectedintheEDservices.
Thiscombinationofdataisurgentlyneededtounderstandtrendsinandrelationshipsbetweenpublic(e.
g.
,citycentre)andprivatespace(e.
g.
,homes)violence.
Together,thesedatawouldenableaneffectivemultiagencyresponsebothnationallyandlocally.
However,eventheexperimentaldatautilisedhereexposessomeimportantgapsinourunderstandingofnighttimeviolence.
Thus,someeventsareviolencepro-moters(suchasStPatrick'sDayandEnglandgamesinthefootballWorldCup),whileothersarenonbelligerent(suchasStGeorge'sDayandtheRugbySixNations)orperhapsevenprotective(forexample,theOlympics).
Therelativeimpactofdifferentholidaysandeventsmayvarywithlo-calityandnation.
ForexampleinCardiff,Wales(whererugbyisoftenconsideredtobethenationalsport),inter-nationalrugbymatchesinvolvingtheWelshteamhavebeenassociatedwithincreasedEDassaultattendances[55].
ResearchisneededtounderstandthefactorsBellisetal.
BMCPublicHealth2012,12:746Page10of13http://www.
biomedcentral.
com/1471-2458/12/746protectingrelativelypeacefulcelebrations,andtherolescommercialisationandlinkagewithalcoholpromotionsplayincouplingcelebrationswithviolence.
Thisstudyhasanumberofimportantlimitations.
EnglishnationalHESA&Edataarestillincomplete.
Althoughafullauditofdataqualityisnotavailable,comparisonwiththeQMAEsuggestsHESA&Erepresent74%ofallpresenta-tionsregardlessofcause[28].
HESA&EthoughcoversallmajorEDswithmuchmissingdataarisingfromotheremergencyserviceproviderssuchaswalkincentres,whichonlyaccountedfor1.
2%ofnighttimeassaultpresentationsinthisdataset.
However,althoughemergencyservicesaretheprincipalresourceforurgentassaulttreatmentatnightnotallindividualsassaulted,orevenrequiringtreatmentwillpresenttothem.
Englandalsohasageneralpracticeoncallsystemwheredoctorscanbeaskedtoattendindivi-duals'placesofresidence.
Further,injuredindividualsmayalsoattempttotreatthemselvesortodelaytreatmentuntilthenextdaywhenthereisagreaterrangeoftreatmentoptionsandtheirattendancetimemayfalloutsideofthisstudy'sinclusioncriteria.
Thestudycannotquantifyhowfrequentlysuchoptionsareexploitedbythoseinjuredinassaults;althoughtheyareunlikelytobeoptionsforthoserequiringimmediateattention.
InEDdata,reasonforattendancewascodedasun-knownin4.
7%ofcasesbutdatacodingreliesonpatientsrevealingthattheirinjurieshavebeensustainedthroughviolenceandthisbeingaccuratelycodedinbusyEDs.
Althoughtheabsenceofanyviolencerelatedpre-sentationsfromnineproviderssuggestsunder-recordingofassaults,currentlyitisnotpossibletoquantifythescaleofsuchmiscodingacrossallEDs.
Theseissueswillinevitablyaffectanycalculationofrates.
However,ourfindingsfocusoncomparativerisks;largelybetweendif-ferentdaysordifferentdemographics.
Wearenotawareofanycalendar,deprivation,orage/sexrelatedbiasinmissingdatathatcouldconfoundourresults,althoughthiscannotbeentirelydiscounted.
OurfocushasbeenonlevelsofemergencypresentationsforassaultandthereforewehaveexcludedEDattendancesforfollowupsrelatingtoapreviousEDattendance.
Wehavenotattemptedtoremovemultiplepresentationsbythesameindividualfordifferentassaults(seemethods).
Conse-quently,demographicanalysesrelatetoprobabilitiesofpresentationsbeingfromaparticulardemographic.
However,acrossthethree-yearstudyperiodonly6.
7%ofindividualspresentedfornighttimeassaultmorethanonceandanalysisofindividuals,ratherthanpresenta-tions,wouldbeunlikelytosubstantivelyaffectresults.
Sportingeventsincludedwereaconvenienceselectionbasedonthosebestknownandhighlypromoted.
Thereareawiderangeofotherlocaleventsthatmighthavebeenincludedinthisanalysisandtheimpactofevenna-tionalevents(suchasafootballcupfinal)mayvarywithlocality;if,forinstance,alocalteamareinvolved.
Theana-lysesundertakenshouldbeconsideredaproofofconceptfortheutilityofEDdata,whichcouldbeimplementedmuchmorewidelywithacompletenationaldataset.
Wecouldnotdistinguishassaultlocations(e.
g.
,homeorcitycentrebar),andthuswehavemadenoassumptionsaboutwhetherassaultstookplaceinpublicplacesorprivateresidences.
WhilethenationalEDsystemdoesnotcur-rentlycollectlocationofassault,thecollectionandshar-ingofsuchdataatlocallevelisincreasing[11,12,57].
Finally,whilethisstudyhasexaminedtheutilityofana-tionalEDdatasetinmeasuringcalendaranddemographicriskfactorsfornighttimeassaultsfurtheranalysesarenowpossible.
EDdataallowsadditionalexplorationoftheresi-denceofthoseinvolvedinviolence(e.
g.
,bypopulationdensity,urbanvs.
rurallocality,etc.
).
Dataonalcoholcon-sumptionbythosepresentingtoEDsisnotcurrentlyavailablenationallybut,routinelycollectedevenfromasubsetofEDs,couldprovideimportantintelligenceontheimpactofalcoholonnightlifeassaults[11].
ConclusionsGlobally,nationalroutinedatacollectionfromEDsisrare.
However,itprovidesnovelintelligenceforpublichealth.
Anationalperspectivehelpsavoiddisplacementissues[58]whenassessingwhetherviolencelevelshavefallenorsimplymovedelsewhere(e.
g.
aneighbouringcity).
EDdataonnighttimeassaultsprovideresidenceinformationandconsequently,measuresofsocio-economicstatus(e.
g.
,IMD)aswellastheabilitytoapplypopulationdenominatorsforidentificationofratesandriskfactors(e.
g.
,byageandsex).
Riskofinvolvementinviolenceisacompositeofatleastenvironment(e.
g.
,citycentremanagement,accesstoalco-hol),otherproximalfactorsspecifictotheindividual(e.
g.
,employmentstatus),andapropensityforviolencethatcanberootedinearlychildhoodexperiences.
Intheserespectsithasthesamecomplexoriginsasothermajorthreatstohealthsuchasobesity[59].
However,untilrecentlytheroleofhealthservicesinthepreventionofviolencehasbeenlargelypassive;withactiveelementslimitedtodealingwiththephysicalandmentalhealthconsequencesofassaultsandabuse.
UseofEDdata,forinstance,hasoftenfocusedonhelpingtargetpoliceandotherregulatoryactivityratherthanbeenconsideredasatooltodirecthealthinterven-tions.
ThisstudyshowshowEDdatamightbeutilisedtoinformfrontlineresponses,includingbyEDsthemselves.
Moreimportantlyhowever,itshouldbecentraltoamultia-gencylifecourseapproachtothepreventionofviolence.
AnationalEDsystemcandescribetheproblem,identifyriskandprotectivefactors,andtargetpreventionandprotectioninterventionsaswellasassesstheirimpact.
Whilecriminaljusticesystemsworktocontainaculturewherecelebra-tions,sportsevents,andholidaysleadtogreaterviolence,Bellisetal.
BMCPublicHealth2012,12:746Page11of13http://www.
biomedcentral.
com/1471-2458/12/746healthservicescouldhelpcreateonewheretheyarenotin-extricablylinked.
AbbreviationsAOR:Adjustedoddsratio;CIs:Confidenceintervals;DRR:Deprivationrateratio;ED:Emergencydepartment;GLM:Generalisedlinearmodelling;HES:HospitalEpisodeStatistics;IMD:IndexofMultipleDeprivation;LR:Logisticregression;LSOA:Lowersuperoutputarea.
CompetinginterestsTheauthorsdeclarethattheyhavenocompetinginterests.
Authors'contributionsMABdesignedthestudyandoversawitsimplementation.
NL,SW,ZQundertookthedataextraction,formattingandqualityassurance.
MAB,KHandNLanalysedthedata.
Allauthorscontributedtothewritingofthemanuscript,reviewedthestudyfindings,readandapprovedthefinalversionbeforesubmission.
Allauthorshadfullaccesstoallofthedata(includingstatisticalreportsandtables)inthestudyandcantakeresponsibilityfortheintegrityofthedataandtheaccuracyofthedataanalysis.
MABisthestudyguarantor.
Allauthorsreadandapprovedthefinalmanuscript.
AcknowledgementsWearegratefultoClarePerkinsandothermembersoftheNorthWestPublicHealthObservatoryfortheirsupportindevelopingthisstudy.
Authordetails1CentreforPublicHealth,LiverpoolJohnMooresUniversity,15-21WebsterStreet,LiverpoolL32ET,UK.
2,CorkUniversityHospital,Wilton,Cork,Ireland.
Received:8June2012Accepted:30August2012Published:6September2012References1.
KrugEG,DahlbergLL,MercyJA,ZwiAB,LozanoR:Worldreportonviolenceandhealth.
Geneva:WorldHealthOrganization;2002.
2.
WorldHealthOrganization:Globalburdenofdisease:diseaseandinjuryregionalmortalityestimatesfor2008;http://www.
who.
int/healthinfo/global_burden_disease/estimates_regional/en/index.
html.
3.
ButchartA,BrownD,Khanh-HuynhA,CorsoP,FlorquinN,MuggahR:Manualforestimatingtheeconomiccostsofinjuriesduetointerpersonalandself-directedviolence.
Geneva:WorldHealthOrganization;2008.
4.
ChaplinR,FlatleyJ,SmithK:CrimeinEnglandandWales2010/11.
London:HomeOffice;2011.
5.
DubourgR,HamedJ:EstimatesoftheeconomicandsocialcostsofcrimeinEnglandandWales:costsofcrimeagainstindividualsandhouseholds,2003/04.
London:HomeOffice;2005.
6.
NHSSecurityManagementServices:CostofviolenceagainstNHSstaff.
London:NHSSecurityManagementServices;2010.
7.
KrugEG:Injurysurveillanceiskeytopreventinginjuries.
Lancet2004,364:1563–1566.
8.
HolderY,PedenM,KrugEG,LundJ,GururajG,KobusingyeO:Injurysurveillanceguidelines.
Geneva:WorldHealthOrganization;2001.
9.
SutherlandI,SivarajasinghamV,ShepherdJ:Recordingofcommunityviolencebymedicalandpoliceservices.
InjPrev2002,8:246–247.
10.
BrinkO:TheepidemiologyofviolenceinDenmark.
JClinForensicMed1998,5:38–44.
11.
QuiggZ,HughesK,BellisMA:Datasharingforprevention:acasestudyinthedevelopmentofacomprehensiveemergencydepartmentinjurysurveillancesystemanditsuseinpreventingviolenceandalcohol-relatedharms.
InjPrev2011,doi:10.
1136/injuryprev-2011-040159.
12.
FlorenceC,ShepherdJ,BrennanI,SimonT:Effectivenessofanonymisedinformationsharinganduseinhealthservice,police,andlocalgovernmentpartnershipforpreventingviolencerelatedinjury:experimentalstudyandtimeseriesanalysis.
BMJ2011,342:d3313.
13.
WardE,DurantT,ThompsonM,GordonG,MitchellW,AshleyD,TheVRISS:Implementingahospital-basedviolence-relatedsurveillancesystem-abackgroundtotheJamaicanexperience.
InjContrSafPromot2002,9:241–247.
14.
SklaverBA,Clavel-ArcasC,Fandino-LosadaA,Guitierrez-MartinezMI,Rocha-CastilloJ,deGarciaSM,Concha-EastmanA:TheestablishmentofinjurysurveillancesystemsinColombia,ElSalvador,andNicaragua(2000–2006).
RevPanamSaludPublica2008,24:379–389.
15.
WahlRA,SiskDJ,BallTM:Clinic-basedscreeningfordomesticviolence:useofachildsafetyquestionnaire.
BMCMed2004,2:25.
16.
EckenrodeJ,CampaM,LuckeyDW,HendersonCRJr,ColeR,KitzmanH,AnsonE,Sidora-ArcoleoK,PowersJ,OldsD:Long-termeffectsofprenatalandinfancynursehomevisitationonthelifecourseofyouths:19-yearfollow-upofarandomizedtrial.
ArchPediatrAdolescMed2010,164:9–15.
17.
WorldHealthOrganization&CentreforPublicHealthLiverpoolJohnMooresUniversity:Violenceprevention:theevidence.
Geneva:WorldHealthOrganization;2010.
18.
ReynoldsAJ,TempleJA,OuS-R,RobertsonDL,MerskyJP,TopitzesJW,NilesMD:Effectsofaschool-based,earlychildhoodinterventiononadulthealthandwell-being.
ArchPediatrAdolescMed2007,161:730–739.
19.
PrinzRJ,SandersMR,ShapiroCJ,WhitakerDJ,LutzkerJR:Population-basedpreventionofchildmaltreatment:theU.
S.
TriplePsystempopulationtrial.
PrevSci2009,10:1–12.
20.
KoeglCJ,FarringtonDP,AugimeriLK,DayDM:Evaluationofatargetedcognitive-behavioralprogramforchildrenwithconductproblems-theSNAPunder12outreachproject:serviceintensity,ageandgendereffectsonshort-andlong-termoutcomes.
ClinChildPsycholPsychiatry2008,13:419–434.
21.
BellisMA,HughesK,AndersonZ,TocqueK,HughesS:ContributionofviolencetohealthinequalitiesinEngland:demographicsandtrendsinemergencyhospitaladmissionsforassault.
JEpidemiolCommunityHealth2008,62:1064–1071.
22.
StucklerD,BasuS,SuhrckeM,CouttsA,McKeeM:ThepublichealtheffectofeconomiccrisesandalternativepolicyresponsesinEurope:anempiricalanalysis.
Lancet2009,9686:315–323.
23.
GaggJ:Effectoftheroyalweddingonemergencydepartmentattendance.
EmergMedJ2011,28:902–903.
24.
EnockKE,JacogsJ:TheOlympicandparalympicgames2012:literaturereviewofthelogisticalplanningandoperationalchallengesforpublichealth.
PublicHealth2008,122:1229–1238.
25.
JonesLA,GoodacreS:Effectof24-halcohollicensingonemergencydepartments:theSouthYorkshireexperience.
EmergMedJ2010,27:688–691.
26.
OwensPL,BarrettML,GibsonTB,AndrewsRM,WeinickRM,MutterRL:EmergencydepartmentcareintheUnitedStates:aprofileofnationaldatasources.
AnnEmergMed2010,56:150–165.
27.
MeerdingWJ,PolinderS,LyonsRA,PetridouET,ToetH,vanBeeckEF,MulderS:Howadequateareemergencydepartmenthomeandleisureinjurysurveillancesystemsforcross-countrycomparisonsinEuropeIntJInjContrSafPromot2010,17:13–22.
28.
NHSInformationCentreforHealthandSocialCare:AccidentandEmergencyAttendancesinEngland(experimentalstatistics).
:;2010–2011.
http://www.
ic.
nhs.
uk/statistics-and-data-collections/hospital-care/accident-and-emergency-hospital-episode-statistics-hes.
29.
TheInformationCentreforHealthandSocialCare.
SecondaryUsesService(SUS);http://www.
ic.
nhs.
uk/services/secondary-uses-service-sus.
30.
TheInformationCentreforHealthandSocialCare.
HESonline;www.
hesonline.
nhs.
uk.
31.
BatesA:MethodologyusedforproducingONS'ssmallareapopulationestimates.
PopulTrends2006,125:30–36.
32.
NobleM,McLennonD,WilkinsonK,WhitworthA,BarnesH,DibbenC:TheEnglishIndicesofDeprivation2007.
London:CommunitiesandLocalGovernment;2008.
33.
BellisMA,HughesK,WoodS,WykeS,PerkinsC:Nationalfive-yearexaminationofinequalitiesandtrendsinemergencyhospitaladmissionforviolenceacrossEngland.
InjPrev2011,17:319–325.
34.
HosmerDW,LemeshowS:Appliedlogisticregression(2ndedition).
Hoboken,NJ:JohnWiley&Sons,Inc;2000.
35.
GagnonDR,Doron-LaMarcaS,BellM,O'FarrellTJ,TaftCT:Poissonregressionformodellingcountandfrequencyoutcomesintraumaresearch.
JTraumaStress2008,21:448–454.
36.
NHSInformationCentreforHealthandSocialCare:TheHESProtocol(June2009).
Leeds:;2009http://www.
hesonline.
nhs.
uk/Ease/servlet/ContentServersiteID=1937&categoryID=331.
37.
NelsonAL,BromleyRDF,ThomasCJ:Identifyingmicro-spatialandtemporalpatternsofviolentcrimeanddisorderintheBritishcitycentre.
ApplGeogr2001,21:249–274.
Bellisetal.
BMCPublicHealth2012,12:746Page12of13http://www.
biomedcentral.
com/1471-2458/12/74638.
CusimanoM,MarshallS,RinnerC,JiangD,ChipmanM:Patternsofurbanviolentinjury:aspatio-temporalanalysis.
PLoSOne2010,5:e8669.
39.
SheenD,TettenbornM:Britishbeer&Pubassociationstatisticalhandbook2010.
London:BritishBeer&PubAssociation;2011.
40.
HorrocksJ,MenclovaAK:Theeffectsofweatheroncrime.
NewZealEconPap2011,45:231–254.
41.
HughesK,BellisMA,CalafatA,JuanM,SchnitzerS,AndersonZ:Predictorsofviolenceinyoungtourists:acomparativestudyofBritish,GermanandSpanishholidaymakers.
EurJPublicHealth2008,18:569–574.
42.
DávalosME,FangH,FrenchMT:Easingthepainofaneconomicdownturn:macroeconomicconditionsandexcessivealcoholconsumption.
HealthEcon2011,doi:10.
1002/hec.
1788.
43.
BachmanJG,O'MalleyPM,ShchulenbergJE,JohnstonLD,BryantAL,MerlineA,BachmanJG,O'MalleyPM,ShchulenbergJE,JohnstonLD,BryantAL,MerlineAC:Thedeclineofsubstanceuseinyoungadulthood:changesinsocialactivities,roles,andbeliefs.
Mahwah,NJ:LawrenceErlbaumAssociates,Inc;2002.
44.
Briceno-LeonR,VillavecesA,Concha-EastmanA:UnderstandingtheunevendistributionoftheincidenceofhomicideinLatinAmerica.
IntJEpidemiol2008,37:751–757.
45.
RiotsCommunitiesandVictimsPanel:Aftertheriots.
London:RiotsCommunitiesandVictimsPanel;2012.
46.
GovernmentHM:Endinggangandyouthviolence:across-governmentreport.
Norwich:TheStationeryOffice;2011.
47.
MatzopoulosR,BowmanB,MathewsS,MyersJ:Applyingupstreaminterventionsforinterpersonalviolenceprevention:anuphillstruggleinlow-tomiddle-incomecontexts.
HealthPolicy2010,97:62–70.
48.
AndaRF,FelittiVJ,BremnerJD,WalkerJD,WhitfieldC,PerryBD,DubeSR,GilesSW:Theenduringeffectsofabuseandrelatedadverseexperiencesinchildhood.
Aconvergenceofevidencefromneurobiologyandepidemiology.
EurArchPsychiatryClinNeurosci2005,256:174–186.
49.
FelittiVJ,AndaRF,NordenbergD,WiliamsonDF,SpitzAM,EdwardsV,KossMP,MarksJS:Relationshipofchildhoodabuseandhouseholddysfunctiontomanyoftheleadingcausesofdeathinadults.
TheAdverseChildhoodExperiences(ACE)Study.
AmJPrevMed1998,14:245–258.
50.
DukeNN,PettingellSL,McMorrisBJ,BorowskyIW:Adolescentviolenceperpetration:associationswithmultipletypeofadversechildhoodexperiences.
Pediatrics2010,125:778–786.
51.
CanadianNurse-FamilyPartnership;http://nfpmcmasterca/indexphp.
52.
AustralianNurse-FamilyPartnershipProgram;http://wwwanfppcomau/.
53.
BarnesJ,BallM,MeadowsP,HowdenB,JacksonA,HendersonJ,NivenL:Thefamily-nursepartnershipprogrammeinEngland:wave1implementationintoddlerhood&acomparisonbetweenwaves1and2aofimplementationinpregnancyandinfancy.
London:DepartmentofHealth;2011.
54.
QuiggZ,HughesK,BellisMA:Effectsofthe2010worldCupfootballtournamentonemergencydepartmentassaultattendancesinEngland.
EurJPublicHealth2012,doi:10.
1093/eurpub/cks098.
55.
SivarajasinghamV,MooreS,ShepherdJ:Winning,losing,andviolence.
InjPrev2005,11:69–70.
56.
HirshonJM,WarnerM,IrvinCB,NixkaRW,AndersenDA,SmithGS,McCaigLF:Researchusingemergencydepartment-relateddatasets:currentstatusandfuturedirections.
AcadEmergMed2009,16:1103–1109.
57.
StephensonJ:DHordersauditofA&Einformationsharingonviolentcrime.
NursTimes2012,108(17):4.
58.
BowersK,JohnsonS,GueretteRT,SummersL,PoyntonS:Spatialdisplacementanddiffusionofbenefitsamonggeographicallyfocusedpolicinginitiatives.
CampbellSystematicReviews2011,3.
doi:10.
4073/csr.
2011.
3.
59.
ParsonsTJ,PowerC,LoganS,SummerbellCD:Childhoodpredictorsofadultobesity:asystematicreview.
IntJObesRelatMatabDisord1999,23(Suppl8):S1–S107.
doi:10.
1186/1471-2458-12-746Citethisarticleas:Bellisetal.
:Nighttimeassaults:usinganationalemergencydepartmentmonitoringsystemtopredictoccurrence,targetpreventionandplanservices.
BMCPublicHealth201212:746.
SubmityournextmanuscripttoBioMedCentralandtakefulladvantageof:ConvenientonlinesubmissionThoroughpeerreviewNospaceconstraintsorcolorgurechargesImmediatepublicationonacceptanceInclusioninPubMed,CAS,ScopusandGoogleScholarResearchwhichisfreelyavailableforredistributionSubmityourmanuscriptatwww.
biomedcentral.
com/submitBellisetal.
BMCPublicHealth2012,12:746Page13of13http://www.
biomedcentral.
com/1471-2458/12/746