8/27/2015

MeasuresofDiseaseFrequency

MeasuresofDiseaseFrequency

Introduction
Forcenturies,knowledgeaboutthecauseofdiseaseandhowtotreatorpreventitwaslimitedbythefact
thatitwasbasedalmostentirelyonanecdotalevidence.Significantadvancesoccurredwhenthestrategyfor
studyingdiseaseshiftedtolookingatgroupsofpeopleandusinganumericapproachtomakecritical
comparisons.

LearningObjectives
Aftersuccessfully
completingthissection,the
studentwillbeableto:
Definewhatismeant
bytheterm
'population'inboth
descriptive
epidemiologyand
analyticepidemiology.
Explainthedifference
betweenfixedversus
dynamicpopulations.
Explainthe
differencesamongthe
parameters:ratio,
proportion,&rate.
Defineandcalculate
prevalence(andbe
abletodistinguish
betweenpointprevalenceandperiodprevalence).Beabletoexplaintheuseofprevalenceinpublic
health.
Defineanddistinguishbetweencumulativeincidenceandincidencerate,anddescribetheirstrengths
andlimitations.
Explaintherelationshipbetweenincidencerateandcumulativeincidence,andbeabletocomputean
estimateofCIfromIR.
Calculatecumulativeincidenceandincidenceratefromrawdataandconvertitintoaformthatenables
youtocomparetheincidenceintwoormoregroups.
Explainwhatismeantbytheterm"atrisk."
Explainwhatismeantby"personyears"ofobservationandbeabletocalculatepersonyearsof
observationfromrawdata.
Explaintheinterrelationshipamongprevalence,incidence,andaveragedurationofdisease(i.e.P=IR
xD).Beabletocalculatetheaveragedurationofdisease,giventheprevalenceandincidencerate.
Explainandcalculate:
cruderates
categoryspecificrates(e.g.genderorrace)
agespecificrates
Beabletodefineandcalculatethefollowingspecialtypesoffrequencymeasurements:
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

1/23

8/27/2015

MeasuresofDiseaseFrequency

morbidityrate
mortalityrate
casefatalityrate
attackrate
livebirthrate
infantmortalityrate
autopsyrate

'Population'
Apopulationissimplyagroupofpeoplewithsomecommoncharacteristic,suchasage,race,gender,or
placeofresidence.A"targetpopulation"isapopulationforwhichyouwouldliketomakesomeconclusions.
Examples:
residentsofMumbai
membersofBlueCross/BlueShield(aU.S.healthinsuranceorganization)
postmenopausalwomeninMassachusetts
coalminersinPennsylvania
malephysiciansintheUnitedStates
membersoftheBUSPHintramuralsoftballteam

FixedversusDynamicPopulations

Fixedpopulation:Inafixedpopulationmembershipisrelativelypermanentandperhapsdefined
bysomeevent.Onceapersonexperiencesthedefiningeventtheyremainpartofthatpopulationas
longastheyarealive.Examplesofrelativelyfixedpopulationsmightinclude:
survivorsoftheatomicblastsinJapan,
veteransoftheVietnamwarortheGulfWars
membersoftheU.S.militarywhosustainedaheadwoundwhilestationedinIraq
residentsofNewOrleanswholosttheirhomesduringhurricaneKatrina.
allbabiesborninTanzaniain2012
Enrollmentinanepidemiologicalstudycanalsobethedefiningeventforapersontoenterafixed
population:
Personswhocompletedandreturnedaquestionnaireinresponsetoaninvitationtojointhe
BlackWomen'sHealthStudy,andwhowerefoundtobeeligiblebystudystaff
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

2/23

8/27/2015

MeasuresofDiseaseFrequency

ResidentsofBostonpublichousingwhometeligibilitycriteria,completedinformedconsent
andabaselinesurvey,andhadonemeetingwithacommunityhealthworkertodiscusssmoking
cessation

Dynamicpopulation:Inadynamicpopulation,membershipisdefinedbycurrentstatus,so
membershipisnotnecessarilypermanent.Apersonisamemberofthepopulationaslongasthey
meetthedefinitionofthepopulation,andtheyceasetobeamembersofthepopulationwhentheyno
longermeetthedefinition.Notethatapersoncanbeamember,leave,andthenbecomeamember
again.Examplesofadynamicpopulationinclude:
residentsofanytownor,state,orcountry
membersofahealthinsuranceplan
womenwhohavegivenbirthwithinthepast12months

Itcanbeabitchallengingattimestodistinguishbetweenfixedanddynamicpopulations,becausethesame
description(e.g.,residentofBoston)canbeinterpretedasaneventoracurrentstate.Therearetwohelpful
solutionstohelpclearupthisconfusion:
1. Askforaclearerdescription.Forexample,compare"everlivedinBostonforatleastoneday"and
"currentlylivesin
Boston".Thefirstdescribesafixedpopulation,thesecondadynamicone.

2. Thinkaboutwhyyouareinterestedinthepopulation.Ifweareinterestedinaquestionthatisonly
relevantwhilethepersonlivesinBoston,suchasriskofaccidentswhileridingabicycle,thenthe
populationisdynamic,becauseonceapersonmovesoutofBostonthereisnoreasontofollowthem.
Ontheotherhand,ifweareinterestedinaquestionthatremainsrelevantevenafterapersonleaves
thecity(e.g.,doesexposuretopollutionleadtolaterdevelopmentofdisease),thenyouwouldwanta
fixedpopulation.

Ratios,Proportions,andRates
Ratio:Aratioisjustanumberthatisobtainedbydividingonenumberbyanother.Aratiodoesn'tnecessarily
implyanyparticularrelationshipbetweenthenumeratorandthedenominator.Forexample,iftherewere100
womeninthisclassand20men,theratioofwomentomenwouldbe100/20or5womenforeachman.This
isjustasimpleratiothatindicateshowmanytimeslargeronequantityiscomparedtotheother.
Proportion:Atypeofratiothatrelatesaparttoawholeoftenexpressedasapercentage(%).For
example,ifthereare120womeninaclassof130students,thentheproportionofwomenis120/130=92%.
Rate:Atypeofratioinwhichthedenominatoralsotakesintoaccountanotherdimension,usuallytime.For
example,speedismeasuredinmiles/houritcanbecalculatedbydividingthenumberofmilestraveledby
thenumberofhoursthatittook.Waterflowmightbequantifiedingallons/minuteonemightmeasurethe
numberofgallonsreleasedduringaperiodoftimeanddividebythenumberofminutesittookinorderto
calculatetheaveragerate.Anexampleofaratethatdoesn'tinvolvetimeismotorvehicledeaths,whichare
oftenreportedasdeaths/vehiclemiles.Thisisonewayinwhichtherelativesafetyofdifferenttypesof
transportation(automobiles,buses,trains,airplanes)canbecompared.
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

3/23

8/27/2015

MeasuresofDiseaseFrequency

Whiletheterm"rate"isusedverybroadlyamongthegeneralpopulation(birthmalformationrate,autopsy
rate,smokingrate,smokingrate,taxrate),inrealityallthesemeasuresareproportions.Forexample,the
smoking"rate"amongadultsisactuallythenumberofadultsinapopulationwhosmokedividedbythetotal
numberofadultsinthepopulationinotherwords,aproportionbecausethenumeratorisasubsetofthe
whole.Onewaytotellaproportionfromatruerateisthataratecanneverbeexpressedasapercentage,
whileaproportionshouldalwaysbeabletobeexpressedasapercentage.

CountsofDiseasedPeople
Countingthepeoplewithdiseaseisanimportantbasicmeasureofdiseasefrequencythatisessentialto
detectingtrendsorthesuddenoccurrenceofaproblem,suchasanepidemic.Simplecountsofthenumber
ofdiseasedpeoplearealsoimportanttopublichealthplannersandpolicymakersforassessingtheneedfor
resourcesinapopulation.

Year

TotalAIDSCasesinCityA

2001

2002

2003

2004

22

2005

75

ThecountofAIDScasesshownhereforCityAwouldlikelystimulatediscussionamongpublicofficials&
healthproviders,butcountdataalonedon'tallowustofullyunderstandtheproblem.Wedon'tknowifallof
thecaseswerelongtimeresidentswhodevelopedAIDSwhilelivinginCityA.Somemayhavemovedinto
townaftertheydevelopedAIDS.Wealsodon'tknowwhetheranyofthecasesmovedawayordied.

Asecondlimitationofjustcountingthenumberofexistingcasesisthatitdoesn'tallowustomakefair
comparisonsofthefrequencyofHIVindifferentcities,sincetheydon'ttakeintoaccountthetotalnumberof
residents.
Whenmeasuringdiseasefrequency,proportionsandratesareveryhelpfulwhencomparinggroups,
becausetheyrelatethenumberofpeoplewithdiseasetothesizeofthepopulationinwhichtheyoccur.
Prevalenceandincidencearethetwofundamentalmeasuresofdiseasefrequency.
Suppose,forexample,thatCityAhad75HIV+residents,whileCityBhad35.Thiswouldsuggestalarger
probleminCityA.

ExistingCases

CityA

75

CityB

35

However,supposeCityAwassubstantiallylarger,with30,000residents,comparedtoonly7,000inCityB.
Tobefair,onewouldneedtotakethisintoaccountbydividingthenumberofcasesineachcitybythe
respectivepopulationsize.

CityA

ExistingCases
75

PopulationSize
30,000

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

Prevalence
0.0025
4/23

8/27/2015

MeasuresofDiseaseFrequency

CityB

35

7,000

0.0050

Inessence,theresultingdecimalfractionsindicatethefrequencyofHIVperpersonineachcity,andwecan
nowseethatCityBactuallyhasahigherprevalenceofHIV+residentsthanCityA,infacttwiceashigh
(0.005vs.0.0025).However,thefrequencyofHIVperindividualisnotaveryintuitiveorusefulconcept.
However,ifwemultiplyeachoftheresultsx10,000,wehavethefrequencyper10,000population.
Obviously,neithercityhasexactly10,000residents,butbyconvertingthedecimalfractionstothisstandard
populationsize,wecannowhaveamoreunderstandabledescriptionoftheprevalenceofHIV+residentsin
eachcity.

ExistingCases PopulationSize

Prevalence

Prevalence

CityA

75

30,000

0.0025

25/10,000

CityB

35

7,000

0.0050

50/10,000

Prevalence
Themeasureofdiseasefrequencywehavecalculatedistheprevalence,thatis,theproportionofthe
populationthathasdiseaseataparticulartime.Prevalenceindicatestheprobabilitythatamemberofthe
populationhasagivenconditionatapointintime.Itis,therefore,awayofassessingtheoverallburdenof
diseaseonthepopulation,soitisausefulmeasureforadministratorswhenassessingtheneedforservices
ortreatmentfacilities.
Epidemiologistssometimesmakeadistinctionbetweenpointprevalence,theproportionofthepopulationat
a'point'intime,andperiodprevalence,theproportionofthepopulationthathasdiseaseoverastatedtime
period.

Example:During1980theFraminghamHeart
Studyexamined2,477subjectsforcataractsand
foundthat310hadthem.So,theprevalencewas
310/2,477=0.125.Thiscanconvenientlybe
expressedas12.5per100or12.5%(percent
means'perhundred').Sincetheexaminationof
thesesubjectstookplaceoverayear,itmightbe
referredtoasaperiodprevalence,andthe
numeratorconceivablycouldincludepeoplewho
hadfirstdevelopedcataractspriorto1980and
peoplewhodevelopedcataractsduring1980just
beforetheirexamwasdone.Notethatallpeople
countedinthenumeratorarealsoincludedinthe
denominator,i.e.,thenumeratorisasubsetofthedenominator.
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

5/23

8/27/2015

MeasuresofDiseaseFrequency

Frequently,thisdistinctionbetweenpointprevalenceandperiodprevalenceisblurry,becauseitisraretobe
abletoassesstheproportionofapopulationthathasadiseaseconditionatexactlythesamepointintime.
Wecouldconsiderourclasstobeapopulation,andIcouldaskthestudentstoraisetheirhandiftheyhadan
upperrespiratorytractinfectiontoday.Icouldeventakeaphotoandusethistovisualizetheprevalenceof
respiratoryinfectionsatthispointintime.So,inthiscasethissnapshotofdiseasefrequencyintheclass
wouldtrulyrepresentprevalenceatapointintime.Inmostcases,however,ittakesmuchlongerthanan
instanttoassesstheproportionofapopulationthatisdiseased.Inotherwords,wehavetobeflexibleinour
definitionofa"point"intime,andwehavetoallowforbroadpoints.Regardless,ofthisdistinctionbetween
pointprevalenceandperiodprevalence,themoreimportantconceptisthatprevalenceisameasureofthe
proportionofthepopulationthathasagivendisease,condition,orcharacteristicatagiventime.Wewillnot
trytodistinguishpointandperiodprevalenceinEP713.

NotesonPrevalence
Notethatwecanalsouseprevalencetoassessthefrequencyofbehaviorsorcharacteristics
thatmightberiskfactorsfordisease.Smokingisn'tadiseaseperseitisariskfactor.However,
itisrelevanttoassesstheprevalenceofthisbehavior.
Also,notethatthe"pointintime"canbeaneventratherthanadistinctcalendartime.For
example,manyelderlymenarefoundtohaveprostatecanceronautopsy,eventhoughthey
wereunawareofitanddiedforotherreasons.Itisappropriatetothinkofthefrequencyof
prostatecanceratthetimeofautopsyasprevalence,eventhoughmenarehavingautopsies
performedatmanydifferentpointsincalendartime.Similarly,militaryrecruitsundergoa
physicalexaminationduringinduction,andtheexamsareperformedatmanydifferenttimes.
Nevertheless,theproportionofinducteesfoundtobecolorblindduringtheirphysicalexamwould
betheprevalenceofcolorblindnessinyoungmen.
Notethatprevalenceisaproportionandnotarate,althoughthelattertermisoftenused.So,
theterms"prevalencerate"and"autopsyrate"aretechnicallyincorrect(althoughcommonly
used).

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

6/23

8/27/2015

MeasuresofDiseaseFrequency

Incidence:Risk,CumulativeIncidence(Incidence
Proportion),andIncidenceRate
Incontrasttoprevalence,incidenceisameasureoftheoccurrenceofnewcasesofdisease(orsomeother
outcome)duringaspanoftime.Therearetworelatedmeasuresthatareusedinthisregard:incidence
proportion(cumulativeincidence)andincidencerate.Ausefulwaytothinkaboutcumulativeincidence
(incidenceproportion)isthatitistheprobabilityofdevelopingdiseaseoverastatedperiodoftimeassuch,it
isanestimateofrisk.KenRothmanusestheexampleofanewspaperarticlethatstatesthatwomenwhoare
60yearsofagehavea2%riskofdyingfromcardiovasculardisease.Aswrittenthisstatementisimpossible
tointerpret,becauseitdoesn'tspecifyatimeperiod.Inordertointerpretriskitisnecessarytoknowthe
lengthoftimethatapplies.A2%riskhasaverydifferentmeaningifitisoverthenext12monthsvs.thenext
10years.Therefore,theincidenceproportion(cumulativeincidence)mustspecifyatimeperiod.For
example,theincidenceproportionofneonatalmortalityisthenumberofdeathsdividedbythenumberof
birthsoverthefirst30daysafterbirth.
Theconceptofriskisfairlyintuitiveifagroupofdiseasefreepeoplewerefollowedforaperiodoftime,one
coulddeterminetheproportionofpeoplewhodevelopedthediseaseatsomepointduringtheobservation
periodinordertoarriveatanestimateoftheprobabilityofdevelopingthatdisease,i.e.therisk.However
appealingthisisforitssimplicity,therearesomedrawbackstothisapproachtoassessingtheoccurrenceof
healthoutcomes,becauseanaccurateassessmentofprobabilityreliesonobservingallsubjectsforthe
entireobservationperiod.Thisisparticularlyaproblemwhenassessinglongtermrisk.
1. First,therearecompetingrisksthatmightresultinthedeathofsomesubjectsbeforetheobservation
periodends,makingitimpossibletoknowwhethertheywouldhavedevelopedtheoutcomeofinterest
iftheyhadnotdiedearlybecauseofanotherrisk.Forexample,studyingtheincidenceproportionof
longtermhealthconditionsamongsoldiersinaconflictzoneiscomplicatedbytheelevatedriskof
dyingincombatbeforetheoutcomecanbeobserved.
2. Asecondproblemisthat,evenifsubjectsdon'tdieforanotherreason,itisdifficulttofollowpeoplefor
longperiodsoftime,andsubjectscanbecomelosttofollowup,whichalsomeansthattheir
outcomestatusisunknown.
3. Athirdproblemisthattheincidenceproportiondoesn'tdistinguishwhenadiseaseoccursaslongasit
iswithinthefollowupperiod.Forexample,ifapopulationisfollowedfor20years,itwouldmakea
differencetothepersonandtotheepidemiologistifthecanceroccurredaftertwoyearsorafter20
years,butbothoftheseoutcomeswouldcountthesamewiththeincidenceproportion.
Forthisreason,theincidenceproportionisgenerallyusedinsituationswherethefollowuptimeisrelatively
shortandthereisrelativelylittlelosstofollowup.Otherwise,epidemiologistsgenerallyusetheincidence
rate.
Ideally,ifwearetoestimateincidence(incidenceproportionorincidencerate),wewouldwanttomeasure
thisinasampleofpeoplewhoaretrulyatriskofdevelopingtheoutcomeofinterest.So,inmeasuring
incidencewewouldliketoexcludeanyonewhowasnotatriskofdevelopingdisease,becausetheyalready
hadthediseaseorbecausetheycouldn'tdevelopit.Forexample,ifonewantedtoestimatetheriskof
developinguterinecancerinpostmenopausalwomen,weideallywouldliketoexcludewomenwhohad
previouslyundergonehysterectomy(removaloftheuterus),sincetheyarenolongeratriskofdevelopingthis
particulartypeofcancer.

DiabetesinaNursingHome
Supposewewereinterestedintheproblemofdiabetesinanursinghomewith800residents.Wewould
beginbydoingbloodtestsonallresidentstodeterminewhichwerediabetic.If50oftheresidentswere
diabeticinitially,thentheprevalenceofdiabetesatthispointintimewouldbe50/800=0.0625.Thestandard
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

7/23

8/27/2015

MeasuresofDiseaseFrequency

wayofexpressingthiswouldbetosaythattheprevalence
was62.5per1000residentsor6.25per100residents,or
0.0625%
Ifwewanttoestimatetheincidenceofdiabetesinthis
populationoverthenext12months,weneedtoexclude
the50peoplewhoarealreadydiabeticandfocusonthe
750residentswhoarediseasefreeinitially.Wewould
thenneedtodoadditionalbloodteststodeterminehow
manynewcasesdevelopedduringthespanoftime.
Becausesomeoftheresidentsmightdieorbetransferred
tootherfacilitiesduringtheyear,weideallywouldliketo
takebloodtestsfrequently,butforfinancialandlogistical
reasons,wemightsimplyconductasecondseriesofbloodtestsafteroneyear.If25werefoundtobe
diabeticattheendofayear,thentheincidencewouldbe25/750=0.0333orabout3.3perhundred(3.3%)
overayear.Notethatwearedescribingthetimespan,i.e.theperiodofobservation,whenwereportthe
incidence.
Whenincidenceisdeterminedinthisway,thatis,byevaluatingthepresenceofdiseaseatthebeginningand
thendividingthenumberofknownnewcasesbythenumberofpeople"atrisk"atthebeginning,itisreferred
toasacumulativeincidenceandcanalsobethoughtofastheincidenceproportion.Whilepeople
commonlyrefertothisasa'rate,'thisisreallyaproportion.Itistheproportionofthe"atrisk"groupthat
developeddiseaseoverastatedblockoftime.
ThecumulativeincidenceofAIDSinMAduring2004:
Cumulativeincidenceiseasytomeasureandiscommonlyusedinawidevarietyofcircumstances.For
example,ifwewantedtodeterminetheincidenceofAIDSinMassachusettsduringcalendaryear2004,it
isn'tfeasibleforustocheckeverycitizenatthebeginningandendoftheyear.Censusdatagivesusarough
ideaofhowmanypeoplelivedinMassachusettsduring2004,andAIDSisareportabledisease,sowecould
gototheMADepartmentofPublicHealthandobtainanestimateofthenumberofpeoplewithAIDSatthe
beginningoftheyear,andwecouldsubtractthisnumberfromthepopulationsizetogetadenominatorthat
representsthenumberofpeople"atrisk"ofdevelopingAIDS.Then,wecouldgobacktoDPHattheendof
thecalendaryearandaskhowmanynewpeoplehadbeenreportedwithAIDS.Thisisournumerator.So,
thecumulativeincidencewouldbe:

ThenumberofnewAIDScasesreportedduringtheyear
ThepopulationofMAatrisk
(i.e.minusexistingcasesatthebeginningoftheyear)
Inreality,therewere523newAIDScasesreportedinMAin2004,andthepopulationwasabout5.7million.
So,thecumulativeincidencewasabout9.2per100,000peopleduring2004.Notethatthedenominatoris
justanestimatebasedonthelastcensus.Inreality,peoplewerebeingaddedtoandsubtractedfromthe
populationcontinuallyasaresultofbirths,deaths,movingintothecity,andmovingout.Wealsodidn'ttake
intoaccountexactlywhentheydevelopedAIDS,althoughweprobablydon'tcarewhethertheydevelopedit
earlierorlaterwithinaoneyearperiod.Nevertheless,thiscumulativeincidenceisausefulnumber,anditis
relativelyeasytogettheinformationweneedtocalculateit.
Itisimportanttospecifythetimeperiodwhenreportingcumulativeincidence.Inthefallsemesterof2003
therewere130studentsinEP713atthebeginningofthesemester,and55ofthemreporteddevelopinga
coldorotherrespiratoryinfectionduringthesemester.So,thecumulativeincidence=55/130=0.42307or
42.3%overthecourseofthesemester.Thetimeperiodofobservationisexpressedinwords.
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

8/23

8/27/2015

MeasuresofDiseaseFrequency

IncidenceRate
Rememberthataratealmostalwayscontainsadimensionoftime.Therefore,theincidencerateisa
measureofthenumberofnewcases("incidence")perunitoftime("rate").Comparethistothecumulative
incidence(incidenceproportion),whichmeasuresthenumberofnewcasesperpersoninthepopulationover
adefinedperiodoftime.Becausestudiesofincidenceinepidemiologyareconductedamonggroupsof
peopleastheymovethroughtime,thedenominatorisactuallyacombinationofthenumberofpeopleand
theamountoftime.Thisisexpressedaspersontime.Thetimeunitscanbeexpressedindays,months,or
years,butshouldbetiedtothelengthofthestudyandaidinterpretationoftheresults.Themostfrequently
encounteredexpressionis"personyears".Thecharacteristicsofcumulativeincidenceandincidencerateare
illustratedintheexamplesbelow.
Note:Whilewegenerallyrefertocumulativeincidence(incidenceproportion)andincidencerateas
measuresofdiseasefrequency,theycanbeappliedtoanysortofoccurrence.Forexample,treatmentsto
cureorrelievediseaseconditionsarealsomeasuredusingtheincidenceproportionorrate,aswewillseein
theexamplebelow.Thekeythingtokeepinmindisthateithermeasureofincidence(unlikeprevalence)
measuresatransitionfromonestatetoanother:welltosick,sicktowell,alivetodead,unborntoborn,etc.

Acomparisonofpainreliefwithtwoanalgesics:
Supposeyouwereaskedtoanalyzethedatafromasmallpreliminaryclinicaltrialwith20subjects.All
subjectshadacomparabledegreeofkneepainfromosteoarthritis,andtheywerebeingcomparedwith
respecttopainreliefafterreceivingastandardpainmedication(DrugB)oranewpainmedication(DrugA).
The20patientswererandomlyassignedtoonedrugortheother,andthereweretensubjectsineachgroup.
Afterreceivingthemedication,theinvestigatorscheckedonthesubjectsathourlyintervalstoseeifthe
subjectshadhadreliefofpain.Foreachsubject,thetimeatwhichpainreliefoccurredwasrecorded.Results
areillustratedinthegraphbelow.

The"X"sindicatewhensubjectsreportedpainrelief.The"O"sattheendindicatesubjectswhodidnotreport
reliefofpain.
Whichgroupappearstohavehadagreaterincidencerateofpainrelief?
Howdidthecumulativeincidenceofpainreliefcompare(theproportionofsubjectsexperiencingpain
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

9/23

8/27/2015

MeasuresofDiseaseFrequency

relief)?

CumulativeIncidence
Sixoftensubjectsineachgroupexperiencedreliefofpain,sothecumulativeincidenceofpainreliefwas
6/10=60%ineachgroup.Whenevercumulativeincidenceisdetermined,onedeterminestheproportionof
subjectswhoexperiencedtheoutcomeofinterestduringablockoftime,withouttakingintoaccountwhen
subjectsdevelopedtheoutcome.Visually,however,itisclearthatifweconsiderwhensubjectsexperienced
relief,theratewasgreaterinthesubjectsreceivingthenewdrug.

IncidenceRate
Inthishypotheticalstudyallsubjectswereobservedforamaximumof10hours,andsomedidnotachieve
painrelief,whileothersgotreliefaftervaryingperiodsoftime.Wecancalculatetheaveragerateofpain
reliefineachgroupbyaddingupthedurationofpainforsubjectsineachgroupanddividingbythenumber
ofsubjectsineachgroup.
Inthegroupreceivingthenewdrugthetimeswere4x1+2+3+4x10=49hoursforthegroup(person
hours).Sotheincidencerateofreliefwas6/49personhoursoronaverage12.2per100personhoursof
observation.Notethatonceasubjectexperiencestheoutcomeofpainrelief,theyarenolongerconsidered
tobeunderobservation.
Inthegroupreceivingtheolddrugthetimeswere4x7+8+9+4x10=85hoursforthegroup(person
hours).Sotheincidencerateofreliefwas6/85personhoursoronaverage7.0per100personhoursof
observation.So,therateofpainreliefwasgreaterinthegroupreceivingthenewdrug.
Whatwehavecalculatedistheincidencerate.Thisisatruerate,becausetimeisanintegralpartofthe
calculation,analogoustomilesperhour(arateofspeed)orgallonsperminute(arateofflow).
Severalthingsarenoteworthyaboutthisincidencerate.
1. Thenumeratoristhesameforbothcumulativeincidenceandincidencerateitisthenumberof
individualswhodevelopedtheoutcomeofinterest(inthiscasepainrelief)duringtheobservation
period.
2. Thedenominatorsforcumulativeincidenceandincidencerateareverydifferent.Forcumulative
incidence,thedenominatoristhetotalnumberof"atrisk"subjectsbeingfollowedforincidencerate,
thedenominatoristhetotalamountoftime"atrisk"ofcontinuedpainforallthesubjectswhowere
beingfollowed.Therefore,wecanonlycalculateanincidencerateifwehaveperiodicfollowup
informationoneachsubject,includingnotonlyiftheydevelopedtheoutcome,butalsowhenthey
developedit.
3. Theincidencerateisamoreaccurateestimateoftherateatwhichtheoutcomedevelops.Cumulative
incidenceisfrequentlyreferredtoasa'rate',butitreallyistheproportionofpeoplewhodevelopthe
outcomeduringafixedblockoftime.Thiswasusefulwhenwewantedtodescribetheincidenceof
AIDSinMassachusetts,becausewedidn'thavedetailedinformationoneachandeveryresidentofthe
state.Wecouldn'ttakeintoaccountwhenpeopledevelopedAIDS.Moreover,wecouldn'taccountfor
peoplewhomovedintothestateinthemiddleoftheyearorpeoplewhomovedoutordied.With
incidencerate,however,wecantakethesefactorsintoaccount.Thestrategyisthesameasinthe
painreliefsampleabove,i.e.thedenominatortakesintoaccountthetotalamountof"atrisk"timefor
thegroup.
4. Incidenceratesareparticularlyadvantageouswhentryingmeasureincidenceinstudieswithdynamic
populationsandinstudieswithfixedpopulationswithrelativelylongfollowuptime.

IncidenceofHIVinaBrothel
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

10/23

8/27/2015

MeasuresofDiseaseFrequency

SupposeyouwantedtostudytheincidenceofbecomingHIV+inagroupof15sexworkersinabrothel.We
beginbytestingthemtomakesuretheyareinitiallyfreeofHIV.FiveofthemalreadywereHIVpositive,so
theywereexcluded,becausetheyarenot"atrisk"ofbecomingHIV+(theyalreadyareHIV+).Weplanto
followthesexworkersforsixyearsandthenendthestudy.WetesttheirHIVstatuseachyearinthe
beginningofJanuary.FourofthembecameHIV+atthetimesshownbythecircled"+"marks.
Thedashedlinesindicatecontinued
followup.Notethatsixofthesex
workersbecome"losttofollowup"as
shownbythe"?"marks.Fiveofthose
whowerelosttofollowupwerestill
HIVnegativewhenwelasttested
them,buttheydisappearedandwe
don'tknowwhatsubsequently
happenedtothem.Subject#10was
foundtobeHIV+inJanuary1990,
andshebecamelosttofollowupin
1994.
Thecumulativeincidencewas
4/10=40%oversixyears,butthis
doesn'ttakeintoaccountthedifferent
amountsoftimecontributedbythose
whodidn'tbecomeHIVpositive,one
ofwhom(Subject#7)wasfollowed
throughoutthesixyearsofthestudy,buttheremainderofwhomwerelosttofollowupsometimebeforethe
endofthestudy(Subjects#2,4,5,6,8).
Theincidencerate,however,cantaketheseproblemsintoaccount,becausethedenominatoristhetotal
"atrisk"observationtimecontributedbyalltensubjects.Thecolumnatthefarrightindicateseachsubject's
"atrisk"observationtime,andthesumforthetensubjectswas26years.So,theIR=4/26personyrs=
0.15/personyear=15/100personyearsofobservation.
NotethatpersontimestoppedbeingcountedassoonasthesubjectwasfoundtobeHIVpositive,because
thesubjectwasnolonger"atrisk"ofdevelopingtheoutcometheyalreadyhadexperiencedit.For
example,Subject#1contributedonepersonyeareventhoughshewasfollowedforallsixyears.

Theincidencerateistypicallyusedin
prospectivecohortstudies(e.g.The
FraminghamHeartStudyorTheNurses
HealthStudy)andrandomizedclinicaltrials
(e.g.ThePhysician'sHealthStudy,which
lookedattheeffectoflowdoseaspirinon
heartdisease).Itismoreaccuratethan
cumulativeincidence,butitrequiresrepeated
followupobservationsoneachsubject,and
studieslikethiscanbeveryexpensiveand
timeconsuming.
Subjectsaresometimesrecruitedintostudies
atdifferenttimes.Eachsubject'sdiseasefree
observationtimeor"atrisk"timecanbe
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

11/23

8/27/2015

MeasuresofDiseaseFrequency

calculatedasthetimefromtheirentryintothestudyuntila)theygetthedisease,b)theybecomelostto
followup,orc)thestudyends.

Example:IncidenceRateintheNurse'sHealthStudyEstrogensand
CoronaryArteryDisease
Here,womenonpostmenopausalhormoneshadan
incidenceof30eventsduring54,308.7personyears
offollowup,or55.2/100,000personyears.

Womenintheuntreatedgrouphad60eventsduring
51,477.5personyearsoffollowupor116.6/
100,000personyears.

AnotherExample:IncidenceRateintheNurse'sHealthStudyObesityand
MyocardialInfarction
Inthisstudy,incidenceratesofMI(myocardial
infarction)werecomparedamong5groupsof
womenbasedontheirbodymassindex(BMI).
Therewerecertainlydifferentnumbersof
womeninthefivegroups,butforeachgroup
theycomputedtheincidenceratebycounting
thenumberwhodevelopedMIanddividingby
thegroup'stotal"atrisk"timeofobservation.
Theresultwasthenconvertedtothenumber
per100,000personyearstofacilitate
comparisonamongthefivegroups.

UnitsforDenominators
Byconvention,allthreemeasuresofdisease
frequency(prevalence,cumulativeincidence,
andincidencerate)areexpressedassome
multipleof10inordertofacilitatecomparisons.Considerthesethreeexamples:
Cumulativeincidence:4/10over6years=0.40=40per100or40%over6years
Incidencerate:3/107.7personyrs=0.02785/personyear=28per1,000personyears
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

12/23

8/27/2015

MeasuresofDiseaseFrequency

Onecanexpressthefinalresultasthenumberofcasesper100people,orper1,000people,orper10,000
people,orper100,000.Generallyoneusesaconvenientmultipleoften.Forexample,theexpressionsbelow
areallequivalent,butthelasttwoarethemostconvenienttotalkabout&thinkabout.Note:Eachtimeyou
movethedecimaltotheright,youincreasethenumberbyafactorof10.

EquivalentExpressionsof
DiseaseFrequency
0.00232newcasesper1personyrs.
0.0232newcasesper10personyrs.
0.232newcasesper100personyrs.
2.32newcasesper1,000person
yrs.
23.2newcasesper10,000person
yrs.
232newcasesper100,000personyrs.

CommonPitfall:Acommonmistakeamongbeginningstudentsisto
failtospecifythedimensionsaftercalculatingincidence,especiallyfor
cumulativeincidence.
IntheexampleforHIVinsexworkers,theincidencerateshouldbereported
as15per100personyears.Notethatthisnumberisnottheequivalentofa
percentage.
Inthesameexample,thecumulativeincidencewas4per10subjects(40%)over6years.Note:You
mustspecifythetimeperiodforcumulativeincidenceoryouwilllosepointsontheexams.

SummaryofBasicMeasuresofDiseaseFrequency
(a
proportion)

Prevalence=#Peoplewithdiseaseatapointintime
#Peopleinthestudypopulationatthatpointintime

CumulativeIncidence=#newcasesofdiseaseinaspecified
period
#Peopleatriskatbeginningofthestudy
period

(a
proportion)

IncidenceRate=#newcasesofdisease
PeopleTime

(arate)

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

13/23

8/27/2015

MeasuresofDiseaseFrequency

RelationshipofIncidenceRatetoCumulativeIncidence
(Risk)
Cumulativeincidence(theproportionofapopulationatriskthatwilldevelopanoutcomeinagivenperiodof
time)providesameasureofrisk,anditisanintuitivewaytothinkaboutpossiblehealthoutcomes.An
incidencerateislessintuitive,becauseitisreallyanestimateoftheinstantaneousrateofdisease,i.e.the
rateatwhichnewcasesareoccurringatanyparticularmoment.Incidencerateisthereforemoreanalogous
tothespeedofacar,whichistypicallyexpressedinmilesperhour.Timehastoelapsetomeasureacar's
speed,butwedon'thavetowaitawholehourwecanglanceatthespeedometertoseetheinstantaneous
rateoftravel.Ratherthanmeasuringriskperse,incidenceratemeasurestherateatwhichnewcasesof
diseaseoccurperunitoftime,andtimeisanintegralpartofthecalculationofincidencerate.Incontrast,
cumulativeincidenceorriskassessestheprobabilityofaneventoccurringduringastatedperiodof
observation.Consequently,itisessentialtodescribetherelevanttimeperiodinwordswhendiscussing
cumulativeincidence(risk),buttimeisnotanintegralpartofthecalculation.Despitethisdistinction,these
twowaysofexpressingincidenceareobviouslyrelated,andincidenceratecanbeusedtoestimate
cumulativeincidence.Atfirstglanceitwouldseemlogicalthat,iftheincidencerateremainedconstantthe
cumulativeincidencewouldbeequaltotheincidenceratetimestime:

CI=IRxT
Thisrelationshipwouldholdtrueifthepopulationwereinfinitelylarge,butinafinitepopulationthis
approximationbecomesincreasinglyinaccurateovertime,becausethesizeofthepopulationatriskdeclines
overtime.Rothmanusestheexampleofapopulationof1,000peoplewhoexperienceamortalityrateof11
deathsper1,000personyearsoveraperiodofyearsinotherwords,therateremainsconstant.The
equationabovewouldleadustobelievethatafter50yearsthecumulativeincidenceofdeathwouldbeCI=
IRXT=11X50=550deathsinapopulationwhichinitiallyhad1,000members.Inreality,therewouldonly
be423deathsafter50years.Theproblemisthattheequationabovefailstotakeintoaccountthefactthat
thesizeofthepopulationatriskdeclinesovertime.Afterthefirstyeartherehavebeen11deaths,andthe
populationnowhasonly989people,not1,000.Asaresult,theequationaboveoverestimatesthecumulative
incidence,becausethereisanexponentialdecayinthepopulationatrisk.Amoreaccuratemathematical
expressionthattakesthisintoaccountis:
CI=1e(IRxT),where'e'=2.71828
Thisconstant'e'arisesinmanymathematicalrelationshipsdescribinggrowthordecayovertime.Ifyouare
usinganExcelspreadsheet,youcouldcalculatetheCIusingtheformula:
CI=1EXP(IRxT)

InthegraphbelowtheupperbluelineshowsthepredictednumberofdeathsusingtheapproximationCI=IR
xT.Thelowerline,inred,showsthemoreaccurateprojectionofcumulativedeathsusingtheexponential
equation.

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

14/23

8/27/2015

MeasuresofDiseaseFrequency

Nevertheless,notethatthepredictionfromCI=IRxTgivesquitereasonableestimatesaslongasthe
cumulativeincidenceremainslessthan10%(equivalentto100deathsinthepopulationof1,000intheabove
graph).

LifeTablesandSurvivalAnalysis
(Optional)
TheequationCI=IRxTprovidesareasonableestimateofriskwhentheincidencerateisrelativelyconstant,
butthisisn'talwaysthecase.Whentheincidenceratechangesovertimethereareotheroptionsfor
estimatingrisk.
1. Onecouldcalculateriskseriallyovershortertimeintervalsduringwhichriskisreasonablyconstant.
However,theintervalshavetobelongenoughtoenablemeaningfulincidenceratesforeachinterval.
2. Anotherapproachthatisusefulwhenriskischangingovertimeistousesurvivalanalysis.Despite
thename,itcanbeusedforanyoutcomeregardlessofwhetheritisfatalornot.Thetablebelow
showsahypotheticallifetablecitedbyRothman.

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

15/23

8/27/2015

MeasuresofDiseaseFrequency

Inthishypotheticalexample,theinitialpopulationatriskwasarbitrarilysetat100,000,andthemortalityrates
ineachgroup(column2,mortalityrates=deathsper100,000personyrs.)wereusedtocalculatethenumber
ofdeathsamongthoseremainingatriskforeachintervalusingtheformulaCI=IRxT.Thus,thefirstage
groupspanned15yearsandthemortalityratewas4.7/100,000personyears,sothenumberofdeathswas
4.7x15=70.5.
Theillustrationbelowshowstheresultsofanalysisofatriallookingattheabilityofzidovudine(ananti
retroviraldrugusedinthetreatmentandpreventionofHIV)toreducematernaltochildtransmission.(The
dataarefromConnorEM,etal.:Reductioninmaternalinfanttransmissionofhumanimmunodeficiencyvirus
type1withzidovudinetreatment.N.Engl.J.Med.1994331:11731180,asquotedinthetextbookby
AschengrauandSeageinTable75,page191inthe2ndedition.)Thiswaspartofprotocol076that
originallydemonstratedtheefficacyofzidovudineinwomenintheUnitedStatesandFrance.Theillustration
belowshowsKaplanMeierplotsoftheprobabilityofHIVtransmissionforthetwogroups.Theestimated
percentagesofinfantsinfectedat72weeksareshownwith95percentconfidenceintervals.Thenumbersof
infantsatriskat24,48,and72weeksareshownbelowthefigure.

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

16/23

8/27/2015

MeasuresofDiseaseFrequency

RelationshipAmongPrevalence,IncidenceRate,and
AverageDurationofDisease
Prevalenceistheproportionofapopulationthathasaconditionataspecifictime,buttheprevalencewillbe
influencedbyboththerateatwhichnewcasesareoccurringandtheaveragedurationofthedisease.
Incidencereflectstherateatwhichnewcasesofdiseasearebeingaddedtothepopulation(andbecoming
prevalentcases).Averagedurationofdiseaseisalsoimportant,becausetheonlywayyoucanstopbeinga
prevalentcaseistobecuredortomoveoutofthepopulationordie.Forexample,aboutadecadeagothe
averagedurationoflungcancerwasaboutsixmonths.Therapywasineffectiveandalmostalllungcancer
casesdied.Fromthetimeofdiagnosis,theaveragesurvivalwasonlyaboutsixmonths.So,theprevalence
oflungcancerwasfairlylow.Incontrast,diabeteshasalongaverageduration,sinceitcan'tbecured,butit
canbecontrolledwithmedications,sotheaveragedurationofdiabetesislong,andtheprevalenceisfairly
high.
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

17/23

8/27/2015

MeasuresofDiseaseFrequency

Ifthepopulationisinitiallyina"steadystate,"meaningthatprevalenceisfairlyconstantandincidenceand
outflow[cureanddeath]areaboutequal),thentherelationshipamongthesethreeparameterscanbe
describedmathematicallyas:
P/(1P)=IRxAvg.Duration,
whereP=proportionofthepopulationwiththediseaseand(1P)istheproportionwithoutit,IRisthe
incidencerate,andAvg.Durationistheaveragetimethatpeoplehavethedisease(fromdiagnosisuntilthey
areeithercuredordie).Ifthefrequencyofdiseaseisrare(i.e.,<10%ofthepopulationhasit),thenthe
relationshipcanbeexpressedasfollow:

Prevalence=(IncidenceRate)x
(AverageDurationofDisease)
Iftheaveragedurationofdiseaseremainsconstant,thenpreventivemeasuresthatreducethe
incidenceofdiseasewouldbeexpectedtoresultinadecreasedprevalence.
Similarly,iftheincidenceremainedconstant,thendevelopingacurewouldreducetheaverage
durationofdisease,andthiswouldalsoreducetheprevalenceofdisease.
Inthelate1990santiretroviraltherapywasintroducedandgreatlyimprovedthesurvivalofpeoplewith
HIV.However,theyweren'tcuredoftheirdisease,meaningthattheaveragedurationofdisease
increased.Asaresult,theprevalenceofHIVincreasedduringthisperiod.
Therelationshipcanbevisualizedbythinkingofinflowandoutflowfromareservoir.Thefullnessofthe
reservoircanbethoughtofasanalogoustoprevalence.Raindropsmightrepresentincidenceortherateat
whichnewcasesofadiseasearebeingaddedtothepopulation,thusbecomingprevalentcases.Wateralso
flowsoutofthereservoir,analogoustoremovalofprevalentcasesbyvirtueofeitherdyingorbeingcuredof
disease.YoucanplaywiththisFlashanimationtoseehowprevalence(heightofthewaterinthereservoir)is
influencedbychangesinincidenceoraveragedurationofdisease.

CalculatingAverageDurationofDisease
Thisrelationshipcanalsobeusedtocalculatetheaveragedurationofdiseaseundersteadystate
circumstances.

IfPrevalence=(Incidence)X(AverageDuration),
thenitfollowsthat

AverageDuration=(Prevalence)/
(Incidence)

Example:Supposetheincidencerateoflungcanceris46newcancersper100,000PY,andthe
prevalenceis23per100,000population,then
AverageDurationofDisease=(23/100,000persons/46/100,000personyears=0.5
year
Conclusion:Individualswithlungcancersurvivedanaverageof6monthsfromthetimeofdiagnosistodeath.
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

18/23

8/27/2015

MeasuresofDiseaseFrequency

SpecialTypesofFrequencyMeasures
Prevalenceandincidencearethefundamentalmeasuresofdiseasefrequency,butspecialnameshave
evolvedforthesemeasures,dependingontheirspecificuse.Allofthesetendtobereferredtoasrates,even
though,strictlyspeaking,theyoftenrefertoproportions(cumulativeincidenceorprevalence).

CategoryspecificRates
Thesearejustprevalenceorincidencemeasuresforspecifiedcategories,suchasage,race,country,
occupation,etc.Forexample,sincediseasefrequencyoftendifferssubstantiallywithage,onefrequently
sees"agespecific"ratesofdisease.

AgeSpecificRatesofStroke
AgeGroup

#NewCasesperYear

GroupSize

CumulativeIncidence
per100,000persons

034

582,083

3544

28

113,561

25

4554

114

114,206

100

5564

320

91,464

350

6574

550

61,155

900

1,126

37,531

3000

75+

Similarly,onemightalsoencounter"racespecific"ratesofdisease,asinthisexample:

Mortalityratesfromselectedcausesfor
whitesandnonwhitesintheUS,1967(rates
per100,000population)

White
Black
Homicide

3.5

32.3

TB

2.5

9.6

21.1

68.6

1.0

3.0

16.6

28.9

7.4

5.5

11.3

5.7

Hypertension
Syphilis
Diabetesmellitus
Leukemia
Suicide

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

19/23

8/27/2015

MeasuresofDiseaseFrequency

InterestingVisualTourofCausesofDeathintheUS
SpecialMeasuresofIncidence
Amorbidityrateistheincidenceofnonfatalcasesofadiseaseinapopulationduringaspecified
timeperiod.Forexample,during1982therewere25,520nonfatalcasesofTBintheUSpopulation.
Themidyearpopulationconsistedof231,534,000.
morbidityrateofTB=25,520/231,534,000=11.0/100,000overoneyear
(Notethatthisisacumulativeincidenceandthereforeisreallyaproportion,notatruerate.)
MortalityRate:In1982therewere1,807deathsfromTBintheUSpopulation,sothemortalityrate
forTBwas7.8permillionoveroneyear(alsoacumulativeincidence,notatruerate).
CaseFatalityRate:thenumberofdeathsfromaspecificdiseasedividedbythetotalnumberofcases
ofthatdisease,i.e.theproportionoffatalcasesofadisease(%).Thisprovidesameasureofthe
severityofthedisease.
Example:ReyesSyndromeisarare,buthighlyfataldiseaseinwhichtheliverandbrainbecome
dysfunctionalduetoabnormalaccumulationofcellularfat.Ittendstooccurwhenpeoplearerecoveringfrom
aviralillness,andittendstobeassociatedwithuseofaspirin,especiallyinchildren.Iftherewere200cases
ofReyessyndromein1982and70died,thenthecasefatalityratewouldbe70/200=35%overoneyear.
[Note:Thisisgenerallycalculatedbydividingthedeathsreportedinagivenyearbythenumberofcases
reportedinthesameyear,butthiscanbemisleadingsincesomediseases(e.g.,TB)aren'trapidlyfatal.
Thus,manyoftheTBfatalitiesthatoccurredin1982wereduetocasesdiagnosedseveralyearsearlier.]
AttackRate:acumulativeincidenceforadiseaseduringaspecificperiod(e.g.,anepidemic).
Example:AfterachurchpicnicinOswego,NYmanyattendeesgotfoodpoisoning.Therewere75peopleat
thepicnic46gotsickwithinseveralhours,sotheattackratewas46/75=61%.
Livebirthrate:thefrequencyoflivebirthsinoneyearper1,000femalesofchildbearingage.
InfantMortalityRate:thefrequencyofdeathsinchildrenunder1yearofageoccurringduringaone
yearperiodper1,000livebirths.

SpecialPrevalenceMeasures
(oftenincorrectlyreferredtoasincidencesorrates).
AutopsyRate:theproportionofpeoplewhohaveaparticularfindingonapostmortemexam(the
prevalenceofacertainfindingamongthepopulationofpeoplewhogetautopsied).
BirthDefectRate:theprevalenceofacongenitalabnormalityatthe"point"ofbirth.Thedenominator
canbeeitherlivebirthsortotalbirths(whichincludeslivebirths+stillbirths),butitgenerallydoesnot
includespontaneouslyabortedfetuses.

TheObstacleCourse

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

20/23

8/27/2015

MeasuresofDiseaseFrequency

UsethisinformationandtheinformationinthetablebelowtoanswertheQuizGroup
questionsbeneaththetable:Asampleof100middleagedandelderlywomenwasfollowedprospectivelyfor
10years.AllsubjectsenteredthestudyonJanuary1,1990andallwomenwerefolloweduntilDecember31,
1999.Nonewerelosttofollowup.Duringthisperiod,5subjectswerediagnosedwithovariancancer,but
theyallsurvivedtotheendofthestudy.Thetimeatwhichthese5subjectsdevelopedcancerisshowninthis
table:
Subject

1990

1991

1992

1993

1994 1995

1996

1997

1998

1999

JWL

cancer

AAB

cancer

PD

cancer

JJ

cancer

RO

cancer

UsethisinformationandtheinformationinthetablebelowtoanswertheQuizGroup
questionsbeneaththetable:

InJanuaryof1990,1,010youngadultsofferedtoparticipateina10yearprospectivestudytodetermine
theirriskofTypeIdiabetes.Thisgroupunderwentaninitialbloodtesttodeterminewhethertheywere
diabetic,andeligiblesubjectswereretestedyearlyforthenext10years.Amongthegroupthatofferedto
jointhestudy:
http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

21/23

8/27/2015

MeasuresofDiseaseFrequency

Therewere2individualswhowerefoundtohavediabetesontheinitialbloodscreeningthese2
peoplewerereferredfortreatmentandwerenotenrolledinthestudy
Therewere1,000whowerediseasefreeandremaineddiseasefreefortheentire10yearsofthe
study
6individualsdevelopeddiabetesduringthecourseofthestudyatthetimesindicatedinthetable
below
2individualswhowereinitiallydiseasefreewerelosttofollowupduringthestudyatthetimes
indicatedinthetablebelow
?=Losttofollowup+=Bloodtestpositivefordiabetes=Continueddiseasefree
followup
Subject#

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

22/23

8/27/2015

MeasuresofDiseaseFrequency

http://sphweb.bumc.bu.edu/otlt/MPHModules/EP/EP713_DiseaseFrequency/EP713_DiseaseFrequency_print.html

23/23