(Causes of Disease)
( ) ( ) ( )
19 19 (Koch ) )
Koch Postulates Robert Koch established the relationship between the tubercule bacillus and tuberculosis. Developed the following three postulates The pathogen occurs in every case of the disease in question associated with it characteristic pathologic clinical manifestations. After being fully isolated from the body and repeatedly grown in pure culture, it can induce disease in a symptom-free host. The pathogen can be recovered from the new host after symptoms appear.
( ) G (triangle mode) 1. 2. (wheel mode) 1. 2. G
(web of causation) (chain of causation)
(necessary) (sufficient) (contributory)
A. M. Lilienfeld
1. 2. 3.
(1) --- --- X === (2) Y--- === ==== X--- === (3) --- ===
( ) ( )
(method of agreement)
(method of difference)
(method of concomitant variation)
Tampon TSS TSS 1979 3 3 7 1980 TSS 12 ( ) 11 1980 CDC 50 40 38 5 13%
Tampon TSS( TSS( ) CDC OR=16 Tampon 70 Tampon(Rely) TSS (1:3) OR 5.67-7.67 Rely 13% 1/3 Rely P&G Rely 1985 FDA Tampon 1990 FDA 1-15g <30
(method of analogy) Burkitt
(method of exclusion) 1972
Strength of Association (1) Rationale Stronger associations are less likely to be due to confounding than are weak associations. The stronger the observed association the more likely is there to be a causal relationship Cigarette smoking and lung cancer
35 1000 RR 84 0.81 0.07 0.90 12.86 126 1.10 0.81 1.13 1.40 508 4.78 4.22 4.87 1.15 779 6.79 6.11 6.89 1.13 1714 15.48 13.25 15.78 1.19
Strength of Association (2) Limitations Not every causal component will be large Passive smoking and lung cancer Non causal associations possible (confounding) Birth order and Down syndrome
Consistency (1) Rationale Repeated observations of an association in different populations under different circumstances increases likelihood of an effect.
Consistency (2) Limitations Lack of consistency does not rule out a causal association because some effects occur only under unusual circumstances Transfusions and HIV Consistency is more relevant when entire causal model is understood (rare).
Specificity (1) Rationale A cause should lead to a single, not multiple, effects A variable is associated with only one disease and a disease associated with only one factor.
Specificity (2) Limitations Factors do have multiple effects and diseases have multiple causes The more diseases that smoking is associated with, the greater the evidence that smoking is not causally associated with any of them
Temporal Relationship Rationale The hypothesized cause must precede the observed effect Mosquito (carrying Plasmodium spp.) bite precedes malaria. Part of the definition of causality Limitations
Dose Response Effect/ Biologic Gradient (1) Rationale The frequency of disease increases with the dose or level of exposure Increased sun exposure leads to increased risk of skin cancer.
10 ( ) <5 5-14 15-24 25-49 >50 7 55 489 475 293 38 61 129 570 431 154 12 RR 1 3.7 7.5 9.6 16.6 27.6
Dose Response Effect/ Biologic Gradient (2) Limitations Absence of dose-response does not rule out causality (threshold effect). The presence of a dose-response effect can be due to relationships other than causal ones (graduated distortion or biases, confounding). Birth order and Down syndrome
Plausibility (Biologic) Rationale Need biologic basis to interpret statistical result. Limitations Dependent upon the scientific assumptions of the time.
Congruence/Coherence (1) Rationale Consistent with existing knowledge of disease process Consumption of uncooked chicken (contaminated by Salmonella), is associated with salmonellosis food poisoning Presence of conflicting information is a source of evidence against causal association
Congruence/Coherence (2) Limitations Absence of coherent information should not be counted as evidence against causality.
Experimental Evidence Rationale Gold standard for evidence of association Randomized trial of Chickenpox vaccine leads to reduction in incidence of chickenpox Limitations Rarely exist for large populations
Analogy (1) Rationale Has an intuitive appeal If drug A (of a certain class) can cause a birth defect then drug B (of the same class) might also cause birth defects Historical observation that vaccination against cowpox prevented smallpox
Analogy (2) Limitations Weak criterion Inventive imagination can come up with an analogy in most situations
1951-1961
1 1959 1960 1961 1961 ( ) 207 8 258 26 5769 349 140 25 60 11 37 2 113 6 30099 5000
2 34(68.0%) 2(2.2%) 36 16 88 104 50 90 140 X 2 =69.40, P<0.001, OR=93.5
3 0-8 (%) 10 14 24 42 51 21,434 21,485 0.24 RR=175, AR=41.76%
1961 12 1962