Specific Death Rates• For example: Early life mortality measures
Specific Death Rates• For example: infant mortality
Specific Death Rates• Neonatal mortality rate(cont’d)• Postneonatal mortality rate
Specific Death Rates• Perinatal mortality rates(cont’d)
Specific Death Rates• Fetal death rate(cont’d)
More Convenient:Summarize an entire situation with a single number calculatedfor each subpopulation, a number that adjusts for differencein compositionTwo Ways:1. Direct method of standardization2. Indirect method of standardization
Direct Method of Standardization:Step 1:Select the standard population.Step 2:compute the expected events that would result if,instead of having different age distributions, all populations wereto have same standard age structureStep 3:Compute the adjusted rate as total expected events in thegroup divided by the total standard population
Adjusted Death RatesAge Adjusted Rates• Direct MethodWhat data available for you for direct method?
Adjusted Death RatesAge Adjusted Rates• Direct Method(cont’d)
Adjusted Death RatesAge Adjusted Rates• Direct Method(cont’d)
Indirect Method of Standardization:Step 1: use a set of standard age-specific rates along with theactual age composition of each populationStep 2: compute the number of events that would have occurredin the two groups if each took on the age specific rates of thestandard population while retaining its own age distributionStep 3: compute standardized event ratio as observed/expectedevents for each groupThe indirect method often concludes with this ratio.Step 4: actual age adjusted rates for each group= event rate inthe standard population x standardized event ratio of the group
Adjusted Death RatesAge Adjusted Rates• Indirect Method Standard Mortality Ratio(cont’d)
• Must know when to use an adjusted rate rather than crude rate• If no confounders: the crude rate is adequate• If confounders present: subgroup specific rates are sufficient• Adjusted rates should be considered if they are meaningful• If distribution of standard population is radically different than thepopulations being compared, standardization is inappropriate• Also, when direct standardization is applied, subgroup specific ratesshould have same general trends in all the groups being compared aswell as in the standard population•Direct method of standardization is used more frequently thanindirect method• Direct method requires subgroup specific rates for all popns.• Application of either method should lead to same conclusion
Life Expectancy• Summary of all age-specific mortalityrates• Estimates hypothetical length of life ofa cohort born in a particular yearThis assumes that current mortality rateswill continue
Sources of Morbidity Statistics• Clinical and hospital• Managed care• Registries• Vital statistics• Surveys• Disease reporting• Insurance and pre-paid med. careplans• Absenteeism records
Terms Related to Morbidity• MorbidityThe extent of illness, injury or disability in adefined population• Incidence of a disease (Incidence rate)The number of new cases of a disease thatoccur during a specified time period(numerator) in a population at risk fordeveloping the disease (denominator)• Prevalence of a disease (Prevalence rate)The number of total cases of disease presentat a particular time (numerator) in a specificpopulation (denominator)• RiskThe likelihood that an individual will contract adisease
CharacteristicsRISK PREVALENCE INCIDENCERATEProbability ofdisease% of pop. withthe diseaseRapidity ofdiseaseoccurrenceNo units No units Cases perperson-timeNewlydiagnosedExisting Newlydiagnosed“Cumulativeincidence”“Incidencedensity”
1. “Do you currently have asthma?”Point prevalence2. “Have you had asthma during thelast 1 years?” Period prevalence3. “Have you ever had asthma?”Cumulative incidence67
Other Measures of morbidity1. Notification rate2. Attendance rates at OPD, healthcenters3. Admission, readmission, anddischarge rates4. Duration of stay in the hospital5. Spells of sickness or absence fromwork or school68
Problems with Numerators• Who has the disease?• Who to include in numerator?• Interview errors
Problems with Denominator• Selective undercounting• Everyone in denominator must havepotential to enter numerator group
Problems with Hospital Data• Selective (many reasons)• Data may be unavailable, etc
IncidenceThe two forms of incidence are:• Cumulative incidence• "risk of disease“• measures the proportion ofpersons who develop a disease in aknown span of time• Incidence rate• "rate of disease“• measures the rate of new diseaseoccurrence over time
Cumulative incidence• Cumulative Incidence =Number of people who get a diseaseduring a specified period * 1000Number of people free of thedisease in the population at risk atthe beginning of a study period73
Incidence Rate• Measures the rapidity with which newlydiagnosed cases of the disease of interestdevelop observe a population count # of new cases measure net time• individuals in population at risk of developing disease• person-time person-years patient-days
Incidence rate per 1,000• Number of new cases of a diseaseoccurring in the population during aspecified period of time * 1000Number of persons who are at riskof developing the disease during thatperiod of time75
Incidence density• if people at risk are observed fordifferent periods of time• The denominator consists of the sumof the units of time that each individualwas at risk and was observed.• This is called person-time and is oftenexpressed in terms of person-monthsor person-years of observation.76
Person time• 1 person at risk who is observed forone year = 1 person-year.• 1 person at risk observed for 5 years= 5 person-years.• 5 people at risk, each of whom isobserved for only 1 year = 5 person-years.77
Incidence Rate (Attack Rate) (cont.)• Can be used for specific exposures• Also used for multiple exposures• Other terms:primary casesecondary attack• secondary cases
Incidence and Attack Rates• Primary Attack rates
Incidence and Attack rates(cont’d)• Secondary Attack rates
Prevalence• Measure of the number (or proportion) ofcases in a given population• What is the difference between prevalenceand incidence?Prevalence → a slice thru a population at agiven point in time that determines who hasthe disease and who does not, while Incidenceonly looks at new cases• In steady state situation (no change in rateor net population)Prevalence = Incidence X Duration of disease
Prevalence• Point prevalence- point in time• Period prevalence- during a definedrange of time
2. Aggregate Measures:Mortality-BasedIndicatorsLife expectancyExpected years of life lostPotential years of life lost
Expectancies and Gaps• From a typical survivalcurve, we can eitherconsider the lifeexpectancy (“E”), or thegap (“G”) betweencurrent life expectancyand some ideal.• Expectancies aregeneric; gaps can bedisease-specific (e.g.,life yrs lost due tocancer)G0%20%40%60%80%100%0 10 20 30 40 50 60 70 80 90 100E
Classifying Health Gaps• Gaps: Compare population health tosome target. = Difference between timelived in health states less than idealhealth, and the specified target• The implied norm or target can bearbitrary, but must be explicit and thesame for all populations being compared.The precise value does not matter
Gaps: Expected Years of LifeLost• Uses population life expectancy at theindividual’s age of deathProblems: different countries may have differentlife expectancies. It’s overall mortality, so cannotidentify impact of a disease.• Standard Expected Years of Life LostReference is to an “ideal” life expectancy• E.g., Japan (82 years for women)• Area between survivorship curve and the chosen norm
Potential Years of Life Lost(PYLL)• PYLL = ( “normal age at death” – actualage at death). Doesn’t much matter whatage is chosen as reference; typically 75• Attempts to represent impact of a diseaseon the population: death at a young age isa greater loss than death of an elderlyperson• Focuses attention on conditions that killyounger people (accidents; cancers)• All-causes or cause-specific
Composite Measures• Aim to represent overall health of a population• Composite measures combine morbidity andmortality into a health index. (An index is anumerical summary of several indicators ofhealth)• Mortality data typically derived from lifetables; morbidity indicators from healthsurveys, e.g.• Self-rated health• Disability or activity limitations• A generic health index
Different Types of Morbidity Scalesfor Use in Composite Measures• Generic instruments cover a wide range ofhealth topics, e.g. reflecting the WHOdefinition. These can be health profiles (e.g.,Sickness Impact Profile, SF-36) or “healthindexes” (e.g., Health Utilities Index,EuroQol)• Specific instrumentsDisease-specific (e.g., Arthritis ImpactMeasurement Scale)Age-specific (e.g., Child Behavior Checklist)Gender-specific (e.g., Women’s HealthQuestionnaire)
Survivorship Functions for HealthStatesG0%20%40%60%80%100%0 10 20 30 40 50 60 70 80 90 100HSurvivorsAgeThis diagram extends the earlierone by recognizing that not allsurvivors are perfectly healthy.The lower area ‘H’ shows theproportion of people in good health(however defined); it shows healthylife expectancy. The top curveshows deaths; intermediate arearepresents levels of disability.Area ‘G’ again represents thehealth gap. The question ariseswhether the people with a disabilityought to be counted with H or withG.Deaths
Health expectancies• Generic term: any expectation of life invarious states of health. Includesother, more specific terms, such asDisability Free Life Expectancy• Two main classes:Dichotomous rating: two health statesHealth state valuations for a range oflevels
I. Dichotomousexpectancies• Here full health is rated 1, and any state ofpoor health (mild, moderate, severedisability) is rated 0.• This leads to Disability-free life expectancy(DFLE): weight of 1 for “no disability” and 0for all other states.• = Expectation of life with no disability, orHealthy Life Expectancy (HLE)• Very sensitive to threshold of disabilitychosenSullivan(1971)
II. Polytomous states and valuations(Wilkins and Adams-1983)• These incorporate many levels of disability intolife expectancy estimates and count time spentwith each level of disability.• Polytomous model (three or more health statesdefined: weights assigned to each; generally 0 to1.0. These may be added together and comparedacross diseases)• = Health-adjusted life expectancy (HALE)• First calculated for Canada by Wilkins. Fourlevels of severity & arbitrary weights.• Recent work uses utility weights. E.g. fromHealth Utilities Index, Quality of Well-Being Scale,EUROQoL, etc.
Polytomous Curves ShowingQuality of SurvivalG0%20%40%60%80%100%0 10 20 30 40 50 60 70 80 90 100HSurvivorsAge(years)This diagram illustratesseveral classes of disability,each having a separateseverity weighting.The area ‘H’ again includeshealthy people, but thedefinition may have changed.The top curve shows deaths;intermediate curves representvarious levels of disability.Deaths
Relationship between Life Expectancy,Health Expectancy and Health-AdjustedLife ExpectancyHealth-AdjustedLife ExpectancyLifeExpectancyHealthyLifeExpectancyBy down-weighting thevarious levels ofdisability,the HALE fallsbetween LE and HLE
Gap Measures: QALYs &DALYs• Gap measures can also use a weighting forintermediate health states. This is necessaryto combine time lost due to ill health withtime lost due to premature mortality• Quality Adjusted Life Years (QALYs) lostCommon outcome measurement in clinical trials,program evaluationRecord extra years of life provided by therapy andquality of that lifeTypically use utility scale running from 0 to 1• DALYS (disability-adjusted life years) lost
Complementarity of HealthExpectancies and Health GapsSLELEHALEHLELE SEYLLSURVIVALHALE HALYPOLYTOMOUSHLE ?DICHOTOMOUSBirthLE = Life Expectancy; SLE = Standard LE; HALE = Health-Adjusted LE;HLE = Healthy LE; SEYLL = Standard Expected Years of Life LostHALY = Health-Adjusted Life Years LostGapsExpectanciesAge
Disability Adjusted LifeYearsPossibilities and Problems
What are DALYs?• DALYs = Disability Adjusted LifeYears• A common measurement unit formorbidity and mortality• Facilitates comparisons of alltypes of health outcomes
Possible use of DALYs• Quantitative analysis of the burdenof disease• Analysis of cost-effectiveness ofalternative interventions• Selection of a package or list ofinterventions deliverable within theavailable budgetJL Bobadilla, WHO: 1996
What is the Global Burden ofDisease study?• Backed by the WHO and the WorldBank• A quantitative overview of the burdenof disease world-wide• Combines information about loss ofquality of life with traditionalepidemiological information onmortality• All health outcomes are expressed inDALYs
Possible use of the GlobalBurden of Disease Study• Epidemiological surveillance oftrends across borders and over time• Projections for future burden ofdisease• Basis of information for decision-making on priorities in healthresearch and health policy
How are DALYsconstructed?• A DALY is a health outcomemeasure with two maincomponentsQuality of life reduced due to adisabilityLifetime lost due to prematuremortality.
DALYs due to living with disability(Red area measures DALYs. Red + white is a “normal”life)82,5 YEARSNODISABILITY
DALYs due to early death(Red area measures DALYs. Red + white is a standardlife)NODISABILITY82,5 YEARS
DALYs due to disability and prematuredeath combined.NODISABILITY82,5 YEARS
Calculation of DALYs(age-weighting and discounting areomitted for didactic reasons)• The calculation of DALYs of a woman who hasbeen deaf since she was 5 and dies when sheis 50: ( Disability weight of deafness is set at0.33) :• Number of healthy life years × the disabilityweight of full health (0) + life years withdisability (50) × disabilty weight for deafness(0,33) + life years lost (30) × the weighting ofdeath (1)• 5 × 0+ 45 × 0,33 + 30 × 1 = 47.35 DALYs
DALYs and QALYs• DALY is a modification of QALY(Quality Adjusted Life Years).• Both concepts combine informationabout length of life and quality of life.• A DALY is a negative QALY.
Relation between QALYs and DALYsDALYs = healthy years lostQALYs = healthy years gainedNODISABILITY82,5 YEARS
How are disability adjustmentsmade?The methods used to assign a disabilityweightings to life years is a critical part ofthe DALY approach.– Diagnostic groups must be chosen anddefined.– Descriptions of those diagnostic groups aredeveloped.– The health states are assigned a disabilityweight to indicate the relative severity ofeach health state.
Current method used forweighting disability• Disability weights are obtained byposing two different Person Trade-Off (PTO) questions to expert panels• PTO1 compares life extensions fordisabled and healthy people• PTO2 compares cures for illnesswith extension of life
Other choices behind DALY• In addition to adjusting the value oflife years with disability weights, andchosing a particular life expectancy,the value of a life year is modified by• Discounting– the value of a life year now is set higherthan the value of future life years• Age weighting– life years of children and old people arecounted less
Calculating DALY score,with age weighting and discounting.• Girl, 5 years old, with below-kneeamputation who lives until she is82,5:• DALYs= life years lived with disease(77,5) × disability weight (0,3) × age-weight (a1)× discounting factor (d2)• 77.5 × 0.3 × a1× d2= 10.5 DALYs
PROBLEMS of the DALYapproach• Is it true?Questions of the validity of theresults• Is it just?Questions of the distributionbetween groups
General problems of validity• What is “Quality of Life” or “Disabilityweighting of life years”?• Can quality of life be measured in a singleand precise number?• Does the same health problem have equalimpact on different persons or groups?• Is there a general agreement to underlyingvalue choices: discounting, age weightingand choice of life expectancy
Validity problems of thecurrent PTO protocol• Lack of simplicity, difficult tounderstand• Forced consistency between twoquestions that are essentiallydifferent• Impossible to answer that allindividuals are equally valuable• The expert panel may not representthe values of other people
Validity problems ofepidemiological estimates• Epidemiological data for Africa, LatinAmerica and Asia are crude estimates.• The uncertainty of the figures ofprevalence, may be hidden in theseemingly mathematical rigor of theresults.• Lack of uniform diagnostic criteria. I.e.what do we mean by “depression”?
Justice• The DALY approach has beencriticised for discriminating– the young– the elderly– future generations (future healthbenefits)– the disabled– women
The young• The 5-year-old girl in the exampleabove yielded 10,5 DALYs.• However, the DALY score withoutage-weight and discounting wouldbe• 77.5 × 0.3 = 23,3 DALYs• This result is twice as high, andwould give her a higher priority.
The elderly• In the literature on justice in healthcare, many agree that given a choice,it is more important to save youngadults than the very old.• This view is captured by the DALY(as a time based measure) itself.
Future generations• The practice of discounting futurebenefits is also controversial.• From society’s viewpoint, why should alife year now be of more value than alife year twenty years ahead?• The implications for preventiveservices versus curative services aresignificant. Preventive interventions aregiven less weight.
The disabled• The DALY approach opens for includingchronic illnesses and disabilities in cost-utility calculation. This is an improvement.• On the other hand, the current person trade-off protocol explicitly assumes that lives ofdisabled people have less value and• implies that disabled people are less entitledto health resources to extend their lives
Example of results• In the protocol behind the present Global Burden ofDisease, a life year for 1000 healthy people has beenset as equally valuable as one life year for– 9524 people with quadriplegia– 2660 blind people– 1686 people withDowns syndrome withoutcardiac malformation– 1499 deaf people– 1236 infertile people• WHO has announced a change in approach.
Women• Underlying value choice: Standardexpectation of life at birth is 82.5 yearsfor women, 80 years for men• The ‘true’ gender gap is greater• Gender gap is adjusted to correspondto ‘biological differences in survivalpotential’• Critique: Might underestimate burdenof disease for females relative to males
Anand S, Hanson K. Disability-adjusted life years: a criticalreview. Journal of Health Economics 1997;16:658-702.Arnesen T, Nord E. The value of DALY life: problems withethics and validity of disability adjusted life years. BMJ1999; 319:1423-1425.Bobadilla J-L, Cowley P, Musgrove P, Saxenian. Design,content and financing of an essential national package ofhealth services. Bulletin of the World Health Organization1994;72:653-662.REFERENCE LIST
Murray CJ, Lopez AD. Global mortality,disability, and the contribution of risk factors:Global Burden of Disease Study. Lancet1997b;349(9063):1436-42.Murray CJ, Lopez AD. Mortality by cause foreight regions of the world: Global Burden ofDisease Study. Lancet 1997c;349(9061):1269-76.Murray CJ, Lopez AD. Regional patterns ofdisability-free life expectancy and disability-adjusted life expectancy: global Burden ofDisease Study. Lancet 1997d;349(9062):1347-52.