Misure di impatto Misure di impatto sulla salute pubblicasulla salute pubblica

• Misure teoriche utili a valutare lMisure teoriche utili a valutare l’’impatto impatto di un programma sanitario sulla salute di di un programma sanitario sulla salute di una collettivitàuna collettività

• Assumono che lAssumono che l’’associazione tra la associazione tra la malattia e lmalattia e l’’esposizione sia causale (non esposizione sia causale (non hanno senso altrimenti)hanno senso altrimenti)

• Basate in genere su misure di Incidenza Basate in genere su misure di Incidenza

Tipi di MisureTipi di Misure

Impatto sugli Impatto sugli espostiesposti

Impatto sulla Impatto sulla popolazionepopolazione

ARAR: rischio attribuibile : rischio attribuibile (Attributable risk)(Attributable risk)

AR%AR%: frazione eziologica : frazione eziologica (Attributable risk fraction)(Attributable risk fraction)

PARPAR: rischio attribuibile nella popolazione : rischio attribuibile nella popolazione (Population Attributable Risk)(Population Attributable Risk)

PAR%PAR%: frazione eziologica nella popolazione: frazione eziologica nella popolazione(Population Attributable Risk Fraction)(Population Attributable Risk Fraction)

Rischio Attribuibile (AR)

II11-I-I00

Qual’è il rischio (incidenza) di malattia attribuibile all’esposizione ??Qual’è l’eccesso di rischio dovuto all’esposizione ??

Incidence

Exposed Unexposed

Iexposti – Inon esposti

Rischio Attribuibile % (AR%) o Frazione Eziologica

Qual’è la proporzione di malattia tra gli esposti che puo: - essere attribuita all’esposizione?? - essere evitata se l’esposizione venisse eliminata

Incidence

Exposed Unexposed

%

exposed

unexposedexposed

I

I - I

(II11-I-I00) / I) / I11

• You are in charge of health prevention• Want to reduce automobile-related deaths• Limited budget

but want to have the greatest impact on reducing deaths

• Cohort study to examine causes for automobile-related deaths:

RR AR AR%

Fast driving 5.0 0.04 80%Drunk driving 10.7 0.136 91%

Relazione tra AR% e RRRelazione tra AR% e RR

Essendo: Essendo:

AR%AR% ==

II11== II00*RR*RR

II0 0 RR - IRR - I00

II0 0 RRRR==

II0 0 (RR - 1) (RR - 1)

II0 0 RRRR==

(RR - 1) (RR - 1)

RRRR

I tribunali americani attribuiscono un evento a un I tribunali americani attribuiscono un evento a un esposizione, se risulta che lesposizione, se risulta che l ’’evento ha almeno una evento ha almeno una

probabilità del 50% di essere dovuto allprobabilità del 50% di essere dovuto all ’’esposizione. esposizione.

AR%AR% ==

2.1 - 1 2.1 - 1

2.12.1== 52.3%52.3%

EsEs: Operaio esposto a cromo esavalente e : Operaio esposto a cromo esavalente e affetto da cancro al polmone. affetto da cancro al polmone.

Cromo Cromo K polmone : RR=2.1 K polmone : RR=2.1

Stima della probabilità Stima della probabilità che K al polmone sia che K al polmone sia

dovuto alldovuto all’’esposizioneesposizione

Prevented fraction

Prevented Fraction (PF)For exposures associated with decreased risk

• If relative risk <1– exposure is protective

• Proportion of potential cases – which would have occurred

if the exposure had been absent

– prevented by the exposure

Prevented fraction

RR -1

I

I -I PF

unexposed

exposed unexposed

Incidence

Unexposed exposed

RR -1

I

I -I

unexposed

exposed unexposed

PF: Vaccine efficacy Pop. Cases Cases/1000 RR

Vaccinated 301,545 150 0.49 0.28

Unvaccinated 298,655 515 1.72 Ref.

Total 600,200 665 1.11

0.72 0.28 - 1

0.72 1.72

0.49 - 1.72 PF

I

I -I

unexposed

exposed unexposed

Expected number of cases among vaccinated if unvaccinated

519 1.72 x 1,000

301,545

Observed number of cases 150

Estimated number of cases prevented 369/519 =72%= PF

Preventable fraction: interpretation

AR e AR% ci informano sull’incidenza o proporzionedi malattia nel gruppo esposto dovuta all’esposizione..

Ma:

Qual è l’eccesso di rischio nella popolazione totale dovuto all’esposizione ???oQual è la percentuale o il numero di casi che potrebberoessere evitati nella popolazione generale se fosse rimossal’esposizione ???

Per rispondere alle precedenti domande è necessarioconoscere la % di esposti della popolazione!!

IIpp

PAR e PAR%PAR e PAR% PARPAR: Incidenza della malattia nella popolazione che : Incidenza della malattia nella popolazione che può essere attribuita allpuò essere attribuita all’’esposizione.esposizione.

PAR%PAR%: Proporzione della malattia nella popolazione che : Proporzione della malattia nella popolazione che può essere attribuita allpuò essere attribuita all’’esposizione.esposizione.

IIpp-I-I00

IIpp-I-I00

IIp p =incidenza nella popolazione generale = p(I=incidenza nella popolazione generale = p(I11)+(1-p) )+(1-p) (I(I00))

p= % di espostip= % di esposti

p(Ip(I11-I-I00))==

Relazione tra PAR% e RRRelazione tra PAR% e RR

p (RR-1)p (RR-1)

p (RR-1) + p (RR-1) + 11

pp = percentuale di soggetti esposti nella = percentuale di soggetti esposti nella popolazionepopolazione

P= % di espostiP= % di esposti

PAR%PAR% = =

Cont. Esempio

• Fast drivers >>>> 20% of the total population

• Drunk drivers>>> 3%of the total population

Dead Not dead Risk

Fast 100 1900 2000 0.050

Slow 80 7920 8000 0.010

180 9820 10000 0.018

Fast driving

44% 100 x 0.018

0.010 - 0.018 PAR%

0.008 0.010 - 0.018 PAR

Dead Not dead Risk

Drunk 45 255 300 0.150

Not d. 135 9565 9700 0.014

180 9820 10000 0.018

PAR%: Drunk driving

22% 100 x 0.018

0.014 - 0.018 PAR%

0.004 0.014 - 0.018 PAR

Conclude

• Driving related deaths in population

– 44% presumably due to fast driving

– 22% presumably due to drunk driving

PAR - How to calculate it?PAR - How to calculate it?

current current poppop

ExposureExposureRemovalRemoval

Implement programImplement program

Disease XDisease X100 cases per 10,000 p-y100 cases per 10,000 p-y

Disease XDisease X60 cases per 10,000 p-y60 cases per 10,000 p-y

PARPAR = = (100-60) cases per 10.000 p-y(100-60) cases per 10.000 p-y40 cases per 10.000 p-y are due to physical inactivity40 cases per 10.000 p-y are due to physical inactivity

PAR%PAR% = (100-60)/10.000 = (100-60)/10.000100/10.000100/10.000

40% of Disease X in this population is due to physical inactivity40% of Disease X in this population is due to physical inactivity

x 100% = 40% x 100% = 40%

EsempioEsempioI consumatori abituali di carni rosse hanno un RR=2 I consumatori abituali di carni rosse hanno un RR=2 di sviluppare cancro al colon.di sviluppare cancro al colon.

Il 25% della popolazione è abituale consumatore di Il 25% della popolazione è abituale consumatore di carni rosse.carni rosse.

LL’’incidenza di cancro al colon nella popolazione incidenza di cancro al colon nella popolazione generale e 50/100.000/anno.generale e 50/100.000/anno.

• Quanti casi su 100.000 persone sono dovuti Quanti casi su 100.000 persone sono dovuti al consumo di carni rosse?al consumo di carni rosse?

• Quale percentuale della malattia può Quale percentuale della malattia può essere attribuita al consumo di carne.essere attribuita al consumo di carne.

Soluzioni Esempio:Soluzioni Esempio:

IIp p = (0.25)(100/100.000/year) + 0.75 = (0.25)(100/100.000/year) + 0.75 (50/100.000/year)(50/100.000/year)

= 62.5/100.000/year= 62.5/100.000/year

PARPAR = I = Ip p - I- I00 = 62.5 - 50.0 = 12.5/100.000/year = 62.5 - 50.0 = 12.5/100.000/year

PAR%PAR% = 12.5/62.5 = 20% = 12.5/62.5 = 20%

PAR - LimitationsPAR - Limitations

• Theoretical conceptTheoretical concept

• Estimates of PEstimates of Pee vary, depending on which vary, depending on which survey data are usedsurvey data are used

• Estimates of RR often come from Estimates of RR often come from different data source than that for Pdifferent data source than that for Pee

• PAR% usually calculated assuming 2 PAR% usually calculated assuming 2 levels only of risk factor, e.g., inactive levels only of risk factor, e.g., inactive versus activeversus active

Population attributable risk (PAR) for polichotomous

exposure:

Pexp(i) * (RRi –1)* 100

1 + Pexp(i) * (RRi –1)

Walter 1976

Example: Passive smoke and cardiovascular disease

PAR% : Pexp(i) * (RRi –1)* 100

1 + Pexp(i) * (RRi –1)

Kawachi, et al., Circulation 1997

Risk ratio for CHD for non-smoking nurses exposed only at work:

Regular 1.91

Occasional 1.58

None 1.00

Percentage of exposure in the population

p(regular)= 16% p(occasional)= 12% p(none)= 72%

PAR% : [.16*(1.91-1)]+[.12*(1.58-1)]

1 + [.16*(1.91-1)]+[.12*(1.58-1)] *100 = 17.7%

The Big Risk NumberThe Big Risk Number

• YouYou’’ve calculated your relative risk and youve calculated your relative risk and you’’ve ve made it statistically significant. Is that enough? Can made it statistically significant. Is that enough? Can you just write up your results, get them published you just write up your results, get them published and start filling out those federal grant applications?and start filling out those federal grant applications?

• You can, but you havenYou can, but you haven’’t yet maximized your t yet maximized your chances for success. Therechances for success. There’’s one last thing to do and s one last thing to do and itit’’s easy as pie. You simply take the innocuous s easy as pie. You simply take the innocuous relative risk number and relative risk number and ““morphmorph”” into a public health into a public health crisis.crisis.

Steven Milloy, The Risky Business of Public Health Research,Steven Milloy, The Risky Business of Public Health Research,SCIENCE WITHOUT SENSESCIENCE WITHOUT SENSE

The Big Risk NumberThe Big Risk Number

• You need to calculate a risk estimate for some You need to calculate a risk estimate for some

population, preferably a large population, population, preferably a large population,

preferably a large population or, better yet, all preferably a large population or, better yet, all

250 million Americans. If you can figure the 250 million Americans. If you can figure the

number of cancer cases or premature deaths number of cancer cases or premature deaths

associated with your risk, youassociated with your risk, you’’re sure to get re sure to get

instant national attention. But how do you do instant national attention. But how do you do

this? Simple. Tell you statisticians you want to this? Simple. Tell you statisticians you want to

calculate an attributable risk. They know how.calculate an attributable risk. They know how.

Steven Milloy, The Risky Business of Public Health Research,Steven Milloy, The Risky Business of Public Health Research,SCIENCE WITHOUT SENSESCIENCE WITHOUT SENSE

The Big Risk NumberThe Big Risk Number• Attributable risk is intended to indicate what Attributable risk is intended to indicate what

percentage of deaths in a population are caused by percentage of deaths in a population are caused by

a risk. For example, saying that a risk. For example, saying that ““16 percent of 16 percent of

deaths are due to being overweightdeaths are due to being overweight”” is an is an

attributable risk. Youattributable risk. You’’ve attributed 16 percent of all ve attributed 16 percent of all

deaths to obesity. All you need to do then is figure deaths to obesity. All you need to do then is figure

out how many deaths there are annually (about 2.2 out how many deaths there are annually (about 2.2

million in the U.S., according to 1991 statistics), million in the U.S., according to 1991 statistics),

then multiply the number of annual deaths by the then multiply the number of annual deaths by the

attributable risk (16 percent). attributable risk (16 percent). Voila! Voila! A public A public

health crisis is born!health crisis is born!Steven Milloy, The Risky Business of Public Health Research,Steven Milloy, The Risky Business of Public Health Research,

SCIENCE WITHOUT SENSESCIENCE WITHOUT SENSE

Examples of Attributable RiskExamples of Attributable Risk

RiskRisk

ObesityObesity

SmokingSmoking

RadonRadon

Chlorinated tap waterChlorinated tap water

Environmental tabacco smokeEnvironmental tabacco smoke

Annual Deaths Attributable to RiskAnnual Deaths Attributable to Risk

350.000 from all causes (Source: derived 350.000 from all causes (Source: derived

from 1995 Harvard University from 1995 Harvard University

Study)Study)

390.000 from all causes (Source: U.S. 390.000 from all causes (Source: U.S.

Surgeon General)Surgeon General)

40.000 from lung cancer (Source: U:S: EPA)40.000 from lung cancer (Source: U:S: EPA)

10.000 from bladder & rectal cancer (Source: 10.000 from bladder & rectal cancer (Source:

Morris et al 1992)Morris et al 1992)

3.000 from lung cancer (Source: U.S. EPA)3.000 from lung cancer (Source: U.S. EPA)Steven Milloy, The Risky Business of Public Health Research,Steven Milloy, The Risky Business of Public Health Research,

SCIENCE WITHOUT SENSESCIENCE WITHOUT SENSE

Examples of Attributable RiskExamples of Attributable Risk

STATISTISTICIANSTATISTISTICIAN’’S WARING: S WARING: ATTIRIBUTABLE RISK MAY NOT ATTIRIBUTABLE RISK MAY NOT

BE SCIENTIFICALLY JUSTIFIABLE. IT IS CALCULATED FROM BE SCIENTIFICALLY JUSTIFIABLE. IT IS CALCULATED FROM

VERY UNCERTAIN STATISTICAL ASSOCIATIONS. THESE VERY UNCERTAIN STATISTICAL ASSOCIATIONS. THESE

ASSOCIATIONS MAY NOT REFLECT TRUE BIOLOGICAL ASSOCIATIONS MAY NOT REFLECT TRUE BIOLOGICAL

CAUSE-AND-EFFECT. AT BEST, A STATISTICAL ASSOCIATION CAUSE-AND-EFFECT. AT BEST, A STATISTICAL ASSOCIATION

IS A REPRESENTATION OF WHAT WAS OBSERVED IN A IS A REPRESENTATION OF WHAT WAS OBSERVED IN A

PARTICULAR POPULATION STUDIED AND IS NOT APPLICABLE PARTICULAR POPULATION STUDIED AND IS NOT APPLICABLE

TO OTHER POPULATIONS NOT STUDIED.TO OTHER POPULATIONS NOT STUDIED.

You, of course, should ignore this warningYou, of course, should ignore this warning..

Steven Milloy, The Risky Business of Public Health Research,Steven Milloy, The Risky Business of Public Health Research,SCIENCE WITHOUT SENSESCIENCE WITHOUT SENSE