1. The objective of this lecture is to have an idea about some advanced subjects in Epidemiology

2. Interaction
When the incidence of a disease in the presence of two or more risk factors differs from the incidence rate expected to result from their individual effect.
Antagonism and synergism :-
Additive
Multiplicative

3. 12 I.R. with presence of factor A 20 I.R. with presence of factor B
1- Additive :-
4 = Usual I.R. of the disease without presence of factor A or B
12 I.R. with presence of factor A
20 I.R. with presence of factor B
12-4 = I.R. because of factor A alone
20-4 = I.R. because of factor B alone
Interaction of factors A+B=4+8+16=28
Attribute Risk
Factor B
Factor A - + 4 12
4+8 20
4+8+16
4+16

4. 2- Multiplicative :-
I.R. in the presence of interaction of factor A & B =
4x3x5x= 60
Factor B
Factor A - + 4 12
3x4 RR 20
3x5x4
5x4 RR

5. Agreement An epidemiological statistical method which shows how much
two investigators agree in their results excluding chance factors

6. 1- Percent agreement : ex
1- Percent agreement : ex. Reading x-ray results by two radiologists, the reading will be either: abnormal, suspected, doubtful or normal:
Examiner I P O N M
Normal L K J I
Doubtful H G F E
Suspected D C B A
Abnormal
Examiner II
Percent Agreement = X 100
A+F+K+P
Total readings

7. 2- In paired observations in which at least one of the findings in each pair was positive,
percent agreement = X100
Observer 1
Positive Negative
Observer 2
positive a b
Negative c d (ignore) a
a+b+c

8. Kappa Statistics : by Cohen 1960
To exclude chance that two unprofessional persons brought from the street and tackle +ve or – ve. The example is CA breast staging by 2 pathologists:
Grading by pathologist A
Grading by pathologist B
Grade II
Grade III
100 60 5 65 4 31 35 64 36
Kappa =
% observed agreement = X 100 = 91%
Percent observed agreement – percent agreement expected by chance alone
100% - percent agreement expected by chance alone
60+31
100

9. So the new expected results (by chance alone)will be :
Kappa : (we have to get % agreement expected by chance that is if pathologist I is working by chance, he will diagnose 60% of the specimens as grade II so if he DX. 60 % of all as stage II (even of the second pathologist) so :
60%x65 =39
60%x35 =21
So the new expected results (by chance alone)will be :
100 21
Grade III 26
Grade II
pathologist B
pathologist A 39 14 38
%agreement = = 53’% expected by chance alone,so
Kappa = = =80%
39 +14
100
91% - 53%
100% - 53% 47

10. Crossover 1- planned : in randomized trials Randomized New Rx. Group 1
Current Rx.
Group 2
Group 1

11. No surgery
2- Unplanned:
Randomized
Surgical
surgery
Refuse surgery
Medical
Require surgery

12. Compliance Non compliance Drop out Drop in
How to assess Compliance : examination, calculate the # of tablets, tests, regularity of visits,….
Treatment of non- compliance : education, rewarding, sometime some sort of punishment.

13. Types Of Controls Hospital based : Easily identified
Aware of antecedent exposure
Subjected to the same intangible selection Factors of the cases
But they are ill

14. 2.Population based:
No selection bias, healthy people, the result could be generalized
Costly, busy people, less motivated, may not recall exposures with the same level of accuracy as the diseased person

15. 3.Special groups: Friends, relatives, neighbors…
Healthy, easy, not costly and cooperative….
Confounding factors related to ethnic background

16. 4.Single controls: ordinary controls, individual or groups
5.Multiple Control: as in CA, leukemia and x-ray exposure
6. Individual Controls: pre and post

17. 7. Matched Pairs:
Controls
Cases
Exposed
Not exposed A B C d

18. 1.Case fatality rate: usually for short term, acute conditions.
Expressing Prognosis
1.Case fatality rate:
usually for short term, acute conditions.
For chronic disease and cancer we use:

19. 2. Five-year survival:
The percent of patients who are alive 5 years after treatment begins or 5 years after diagnosis.
Nothing magical about the number (5) i.e. it could be 7 or 10 and no biological changes occur during this period to justify it’s use but they see that usually most of deaths from cancer occur within this period so it is used as an index of success in cancer treatment.
Two problems:
- Lead time bias
- 5 years of observation is necessary

20. Number of alive at the end of the year
3.Observed survival:
Year of treatment
Number of Pat. treated
Number of alive at the end of the year
1st.
2nd.
3rd.
4th
5th.
2008 92 48 22 11 8 7
2009 88 42 18 10 6
2010 76 36 16 -
2011 98 50 23
2012 84 40
Total
438
176 79 32 14

21. P1= 176 x100
438
P2= x100
176-40
P3= x100
79-23
P4= x100
32-11
P5= x100
14-6

22. Probability of surviving 1year = P1
Probability of surviving 2year = P1 x P2
Probability of surviving 3year = P1 x P2 x P3
Probability of surviving 4year = P1 x P2 x P3 x P4
Probability of surviving 5year = P1 x P2 x P3 x P4 x P5

23. 4. Median survival time : Why we use MST?
The length of time that half of the study population survives.
Why we use MST?
Less affected by extremes than the mean
If we use the mean we have to wait till all the patients die.

24. 5. Relative survival rate:
Observed survival in people with the disease
Expected survival if the disease is absent