1. Measure of impact FETP India
What comes from what?
Measure of impact
FETP India

2. Competency to be gained from this lecture
Calculate an attributable fraction

3. Terminology Attributable fraction Etiologic fraction Attributable risk
Same thing

4. True or false?
The relative risk of lung cancer among smokers is 9
Therefore, if nobody was smoking, the incidence of lung cancer would be divided by nine
 False Measures of association are not measures of impact: It is not only the strength of association that matters but also the proportion of the population exposed to the risk factor.

5. True or false?
90% of patients with lung cancer are smokers
Therefore, if nobody was smoking, the incidence of lung cancer would be reduced by 90%
 False The proportion of a disease that may be explained by a specific exposure does not depend on the proportion of cases exposed. It also depends upon the strength of the association (90% of patients with lung cancer also eat masala dosa)

6. Presentation of the data of an analytical study in a 2 x 2 table
Ill Non ill Total
Exposed a b a+b
Non exposed c d c+d
Total a+c b+d a+b+c+d
Rappelez vous de ce tableau à 4 cases

7. Key areas Attributable fraction among exposed
Attributable fraction in the population

8. Attributable fraction among exposed
Proportion of disease that could be prevented if the risk factor was suppressed from among a population where 100% are exposed
This does not make sense in the absence of causality
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

9. Concept of attributable fraction among exposed
Let’s see what is the risk of disease among exposed…
Of this risk, what is the proportion that is attributable to the exposure?
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

10. Attributable fraction among exposed: From concept to formula
Estimate the risk among exposed (Denominator)
Estimate the proportion of it that is attributable to the exposure (Numerator)
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

11. Attributable fraction among exposed: Finding the right formula
Portion of the risk among exposed that is attributable to the exposure
_________________ Risk among exposed
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

12. Attributable fraction among exposed
The risk difference quantifies the excess of risk among exposed in a cohort
Ill Non ill Total
Exposed a b L1
Non exposed c d L0
Total a+c b+d L1 + L0
Rappelez vous de ce tableau à 4 cases
Risk difference = R1- R0 = (a/L1) - (c/L0)
Attributable fraction among exposed

13. Attributable fraction among exposed
Risk difference
_________________ Risk among exposed
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

14. Attributable fraction among exposed in a cohort study: Conceptual formula
Ill Non ill Total
Exposed a b L1
Non exposed c d L0
Total a+c b+d L1 + L0
Rappelez vous de ce tableau à 4 cases
AF e = (R1-R0) /R1 = [(a/L1) - (c/L0)] / (a/L1)
Attributable fraction among exposed

15. Attributable fraction among exposed: Formula used in calculations
AF e = (R1-R0) /R1 = RR-1 / RR
Rappelez vous de ce tableau à 4 cases
(Can be demonstrated)
Attributable fraction among exposed

16. Attributable fraction among exposed
Risk of leishmaniasis according to water bodies within 25 metres of house, Chatrakhali, West Bengal,
Leishmaniasis Non ill Total
Water bodies
No water bodies
Total
Exemple
Quelle est la population?
Quelle est l'exposition?
Prière de vérifier les calculs!
Traduction en langage naturel: qui se lance?
Relative risk = (139/626) / (11/125) = 22% / 9% = 2.5
AF e = (2.5-1)/2.5 = 0.6 = 60% (Try to calculate using both formula)
Attributable fraction among exposed

17. Attributable fraction among exposed
The attributable fraction among exposed only depends on the relative risk
RR= 1; AFe = 0
RR= 2; AFe = 1/2 = 50%
RR= 3; AFe = 2/3 = 66%
RR= 4; AFe = 3/4 = 75%
RR= 5; AFe= 4/5 = 80%
And so on… with a plateau
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

18. Attributable fraction among exposed
Population attributable fraction among exposed according to the relative risk
100.0%
90.0%
AFe
80.0%
70.0%
60.0%
Attributable fraction among exposed
50.0%
40.0%
30.0%
20.0%
10.0%
0.0% 1 2 3 4 5 6 7 8 9 10
Relative risks

19. Case of a protective exposure: Preventable fraction among exposed
Ill Non ill Total
Exposed a b L1
Non exposed c d L0
Total ? ? L1 + L0
Lien avec les types d'études
Sens de l'observation
On ne connaît pas le nombre de malades et de non malades au début de l'étude
On connaît le nombre des exposés et des non exposés
PFe = (R0-R1) /R0 = 1 - RR
Attributable fraction among exposed

20. Vaccine efficacy = Preventable fraction among exposed to a vaccine
Proportion of cases potentially avoided among vaccinated
Proportion of the cases that would have occurred among vaccinated but that were avoided because of vaccination
VE = (ARNV - ARV) / ARNV
Nous allons voir les mesures dans trois contextes différents
Attributable fraction among exposed

21. Attributable fraction among exposed: Case-control study
Cases Controls Total
Exposed a b ?
Unexposed c d ?
Total C1 C0 C1+C0
Etude cas- témoins correspondant à la cohorte
Rétrospective dans sa logique
Sac de cas contre sac de témoins
Sens de la lecture
f1 et f2
Adieu risque
If rare disease: (OR - 1)/OR
Vaccine efficacy: 1 - OR
Attributable fraction among exposed

22. Attributable fraction among exposed
Habits of sleeping in work clothes among scrub typhus cases and controls, Darjeeling, West Bengal, India,
S. typhus Controls Total
Sleeping in work clothes
Changing clothes to sleep
Total
Exemple
Concept OR
Signification
Incidence: 4.1%
Rare
RR estimé
Attention ad/bc
Odds ratio = (66x33)/(56x13) =3.0
AF e = (3-1)/3 = 66%
Attributable fraction among exposed

23. Attributable fraction among exposed
Number of hepatitis B vaccine doses received by HBsAg+ cases and controls, Romania,
HBsAg (+) HBsAg (-)
Cases Controls
3 doses 3 25
< 3 doses 4 2
7 27
OR= 0.06 Vaccine efficacy = 1- OR = 94%
Attributable fraction among exposed

24. Key areas Attributable fraction among exposed
Attributable fraction in the population

25. Attributable fraction in the population
Proportion of disease that could be prevented if the risk factor was suppressed in the population
Does not make sense in the absence of causality
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

26. Concept of attributable fraction in the population
Let’s see what is the risk of disease in the population…
Of this risk, what is the proportion that is attributable to the exposure?
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

27. Attributable fraction in the population: From concept to formula
Estimate the risk in the population (Denominator)
Estimate the proportion of it that is attributable to the exposure (Numerator)
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

28. Attributable fraction in the population: Finding the right formula
Portion of the risk in the population that is attributable to the exposure
_________________ Risk in the population
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

29. Excess of risk in the population in a cohort
Ill Non ill Total
Exposed a b L1
Non exposed c d L0
Total a+c b+d L1 + L0
Rappelez vous de ce tableau à 4 cases
Excess of risk in the population = Rp- R0 = (a+c/L1+L0) - (c/L0)
Attributable fraction in the population

30. Attributable fraction in the population
Excess of risk in the population
_________________ Risk in the population
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

31. Attributable fraction in the population
Attributable fraction in the population in a cohort study: Conceptual formula
Ill Non ill Total
Exposed a b L1
Non exposed c d L0
Total a+c b+d L1 + L0
Rappelez vous de ce tableau à 4 cases
AFp = (Rp-R0) /Rp
Attributable fraction in the population

32. AFp = (Rp-R0) /Rp = P1x(RR-1/RR) P1: Proportion of cases exposed
Attributable fraction in the population in a cohort study: Formula used in calculations
AFp = (Rp-R0) /Rp = P1x(RR-1/RR)
P1: Proportion of cases exposed
Rappelez vous de ce tableau à 4 cases
(Can be demonstrated)
Attributable fraction in the population

33. Attributable fraction in the population
Risk of leishmaniasis according to water bodies within 25 metres of house, Chatrakhali, West Bengal,
Leishmaniasis Non ill Total
Water bodies
No water bodies
Total
Exemple
Quelle est la population?
Quelle est l'exposition?
Prière de vérifier les calculs!
Traduction en langage naturel: qui se lance?
Relative risk = (139/626) / (11/125) = 22% / 9% = 2.5
AF e = (2.5-1)/2.5 = 0.6 = 60%
AF p = (139/150) x 60% = 93% x 60% = 56%
(Try the two formulas)
Attributable fraction in the population

34. Attributable fraction in the population
Incidence of gastro-enteritis according to food items consumed, wedding dinner, West Bengal, 2005 In c id e nc a m o n g t h os wh
ate d i ot at I l
Tot %
Cold d r
inks
153
186
82% 25 41
61%
Sal
144
189
76% 34 38
89% Fi s
h Fry
173
195 5 32
16%
Fish C
urry
119
149
80% 59 78 Mu
ton
158
200
79% 20 27
74% R
osgulla
159
202
79% 19 25
76%
Ice cre a m
166
211
79% 12 16
75%
Pan
148
181
82% 30 46
65%
Relative risk: 5.6, Attributable fraction among exposed = 82%
Attributable fraction in the population = (173/178) x 82% = 80%
The source identified explains most cases
Attributable fraction in the population

35. Attack rate of gastro-enteritis by food items, Coochbehar, West Bengal, India, 2005
Relative risk
95% confidence interval
Ate
Did not eat
Raw custard*
90%
51%
1.8
Fruits
76%
86%
0.88
Sugar candy
75%
88%
0.85
Puffed rice
80%
84%
0.95
Seasoned rice
67%
0.79
This slide summarizes the attack rate by food items. You can see that of five food items, the raw custard was the only food item for which the attack rate was significantly different among cases and controls, with attack rates of 90% and 51% respectively among those who ate and those who did not eat. However, you can see that the relative risk was not high and than the population attributable fraction suggested that this food item did not explain the majority of cases.
Thus, we had to look further.
* 87% of cases exposed, population attributable fraction: 38%

36. Additional investigations, gastroenteritis outbreak, Coochbehar, West Bengal, India, 2005
Raw custard not refrigerated contained raw milk, flour, sugar, honey, banana, clarified butter
In addition to food items, all persons took a spoon of diluted raw milk, explaining:
High attack rate in persons unexposed to custard
Low relative risk / attributable fraction for raw custard
The attributable fraction in the population pointed to the fact the source of infection had two vehicles
The raw custard had been prepared at 2 PM, and was not refrigerated. It was made of a mixture of raw milk, flour, sugar, honey, banana and clarified butter.
The raw milk was purchased from the same salesman in the two houses. It was the only ingredient common to the two festivals.
We also learned that, in addition to these food items, all persons took a spoon of diluted raw milk, explaining the high attack rate in persons unexposed to the custard and the low relative risk / attributable fraction for the raw custard.
From the lab point of view, the rectal swab only grew proteus sp, which was most likely of no importance. No food leftovers available for testing.

37. Attributable fraction in the population
Depends upon the relative risk
Depends upon the proportion of cases exposed
Can never exceed the proportion of cases exposed
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

38. Attributable fraction in the population
From the proportion of cases exposed to the attributable fraction in the population
20% of cases exposed, OR very high
Attributable fraction close to 20%
100% of cases exposed, OR = 1
Attributable fraction = 0%
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

39. Population attributable fraction
Population attributable fraction according to the relative risk for various level of exposure frequency among cases
100.0%
Pe 10%
90.0%
Pe 25%
Pe 50%
80.0%
Pe 75%
Pe 100% (AFe)
70.0%
60.0%
Population attributable fraction
50.0%
40.0%
30.0%
20.0%
10.0%
0.0% 1 2 3 4 5 6 7 8 9 10
Relative risks

40. Attributable fraction in the population: Case-control study
Cases Controls Total
Exposed a b ?
Unexposed c d ?
Total C1 C0 C1+C0
If rare disease: P1x [(OR - 1)/OR]
Etude cas- témoins correspondant à la cohorte
Rétrospective dans sa logique
Sac de cas contre sac de témoins
Sens de la lecture
f1 et f2
Adieu risque
Attributable fraction in the population

41. Attributable fraction in the population
Sleeping in work clothes and scrub typhus, Darjeeling, West Bengal, India,
S. typhus Controls Total
Sleeping in work clothes
Changing clothes to sleep
Total
Exemple
Concept OR
Signification
Incidence: 4.1%
Rare
RR estimé
Attention ad/bc
Odds ratio = (66x33)/(56x13) =3.0
AF e = (3-1)/3 = 66%
AF p = (66/122) x 66% = 54% x 66% = 36%
Attributable fraction in the population

42. Attributable fraction in the population
Characteristics of acute hepatitis B cases and controls, Karachi, Pakistan, 2001
Exposure
Acute hepatitis B (N=67)
Healthy controls
(N=247) OR
Injection
43 (?%)
68 (?%) ?
Transfusion
2 (?%)
1 (?%)
Fraction attributable to injections: (43/67) x (4.7-1/4.7) = 64% x 79% = 50%
Fraction attributable to transfusion: (2/67) x (7.6-1/7.6) = 3% x 87% = 2.5%
Attributable fraction in the population

43. Risk assessment: Rule of thumb
There is more death and disability from frequent exposure to lower risks than to rare exposures to higher risks
Examples:
More people die from marginally elevated blood pressure (common) than from frankly elevated blood pressure (uncommon)
More people acquire HCV from unsafe injection (common exposure, lower risk) than from unsafe blood (rare exposure, high risk)
Nous allons voir les mesures dans trois contextes différents
Attributable fraction in the population

44. Measures of impact in population for protective exposures
If you want to know what was actually prevented by the prevention measure as used currently
Calculate the fraction prevented in the population
Proportion of the population USING the prevention measure x preventable fraction among exposed
If you want to know how much could be prevented if the prevention measure was used in the whole population
Consider “failure to use the prevention measure” as risk factor (1/RR or 1/OR)
Calculate attributable fraction in the population of the “failure to use the prevention measure”
Proportion of the population NOT USING the prevention measure x attributable fraction among exposed

45. Attributable fraction in the population
Take home messages
The attributable fraction among exposed translates the strength of association into an estimate of the proportion of cases attributable to the exposure among exposed
The attributable fraction in the population translates the strength of association and the proportion of cases exposed into an estimate of the proportion of cases attributable to the exposure in the population
Attributable fraction in the population