Association & causation

Epidemiology ,association and causation, exposure-outcome relationship

1. ASSOCIATION & CAUSATION Dr. Priyanka Sharma III year M.D.S DEPARTMENT OF PUBLIC HEALTH DENTISTRY JSS DENTAL COLLEGE & HOSPITAL 1

2. CONTENTS INTRODUCTION APPROACHES FOR STUDYING DISEASE ETIOLOGY HISTORY WHAT IS ASSOCIATION TYPES OF ASSOCIATION WHAT IS CAUSE GENERAL MODELS OF CAUSATION TYPES OF CAUSAL RELATIONSHIP CRITERIA FOR A CAUSAL RELATIONSHIP GUIDELINES FOR JUDGING WHETHER THE ASSOCIATION IS CAUSAL EVIDENCE FOR A CAUSAL RELATIONSHIP DERIVING CAUSAL INFERENCES: EXAMPLE MODIFIED GUIDELINES FOR EVALUATING THE EVIDENCE OF A CAUSAL RELATIONSHIP MEASURES OF ASSOCIATION CONCLUSION REFERENCES 2

3. INTRODUCTION 3

4. In The Magic Years , Fraiberg (1959) characterized every toddler as a scientist, busily fulfilling an earnest mission to develop a logical structure for the strange objects and events that make up the world that he or she inhabits . Each person develops and tests an inventory of causal explanations that brings meaning to the events that are perceived and ultimately leads to increasing power to control those events . The fruit of such scientific labours is a working knowledge of the essential system of causal relations that enables each of us to navigate our complex world. 4

5. In epidemiological studies, a scertainment of cause-effect relationships is one of the central and most difficult tasks of all scientific activities. Epidemiological principles stand on two basic assumptions : Human disease does not occur at random . The disease and its cause as well as preventive factors can be identified by a thorough investigation of population. Hence, identification of causal relationship between a disease and suspected risk factors forms part of epidemiological research. 5

6. APPROACHES FOR STUDYING DISEASE ETIOLOGY 6

7. Strength of evidence of studies Systematic review or meta-analysis of RCTs Double-blind RCTs Single-blind RCTs Randomized , controlled trials (RCTs) Non-randomized / uncontrolled experimental studies cohort studies Case-control studies Ecological studies Cross-sectional studies Expert opinions, anecdotal reports 7 Approach for studying disease etiology

8. Conceptually, a two-step process is followed in carrying out studies and evaluating evidence: 1.  Determine whether there is an association between an exposure or characteristic and the risk of a disease. To do so, we use:    a. Studies of group characteristics : ecologic studies    b. Studies of individual characteristics : case-control and cohort studies    2. If an association is demonstrated, we determine whether the observed association is likely to be a causal one or not. 8

9. Ecologic Studies The first approach in determining whether an association exists might be to conduct studies of group characteristics, called ecologic studies. ECOLOGICAL FALLACY : Eg.relationship between breast cancer incidence and average dietary fat consumption in each country ECOLOGICAL INFERENCE FALLACY: Eg.areas with high concentrations of farm animals are also the areas with lowest concentrations of childhood asthma. It’s a fallacy to then assume that a child who has asthma must not live near any farm animals 9

10. So? Do You Have Enough Info To Inform The Patient? 10

11. Recognizing the limitations discussed above of ecologic studies that use only group data , we turn next to studies of individual characteristics: case-control and cohort studies . In case-control or cohort studies , for each subject we have information on both exposure (whether or not and, often, how much exposure occurred) and disease outcome (whether or not the person developed the disease in question). 11

12. HISTORY 12

13. Historical Theories of disease causation “ Supernatural causes”& Karma Theory of humors (humor means fluid) The miasmatic theory of disease Theory of contagion Germ theory Koch’s postulates 13

14. EVIDENCE FOR A CAUSAL RELATIONSHIP In 1840, Henle proposed postulates for causation that were expanded by Koch in the 1880s.The postulates for causation were as follows:    1.    The organism is always found with the disease.    2.    The organism is not found with any other disease.    3.    The organism, isolated from one who has the disease, and cultured through several generations, produces the disease (in experimental animals). Koch added that “Even when an infectious disease cannot be transmitted to animals, the ‘regular’ and ‘exclusive’ presence of the organism [postulates 1 and 2] proves a causal relationship.” 14

15. These postulates, though not perfect, proved very useful for infectious diseases However, as apparently noninfectious diseases assumed increasing importance toward the middle of the 20th century, The issue arose as to what would represent strong evidence of causation in diseases that were generally not of infectious origin. 15

16. ASSOCIATION 16

17. Association Syn : Correlation, Covariation, Statistical dependence, Relationship Defined as occurrence of two variables more often than would be expected by chance. An association is present if probability of occurrence of a variable depends upon one or more variable . ( A dictionary of Epidemiology by John M. Last) 17

18. If two attributes say A and B are found to co-exit more often than an ordinary chance. It is useful to consider the concept of correlation. Correlation indicates the degree of association between two variables Causal association: when cause and effect relation is seen. 18

19. Pyramid Of Associations 19 Raj Bhopal : Cause and effect: the epidemiological approach Causal Non-causal Confounded Spurious Positive /negative

20. Positive: Occurrence of higher value of a predictor variable is associated with occurrence of higher value of another dependent variable. Ex - education and suicide. Negative: Occurrence of higher value of a predictor variable is associated with lower value of another dependent variable . Ex - Female literacy and IMR 20

21. Causal : Independent variable must cause change in dependent variable. Definite condition of causal associations are time and direction Ex – salt intake and hypertension Non-causal : Non-directional association between two variables. Ex – alcohol use and smoking 21

22. Spurious Association (Spurious= not real, artificial, fortuitous, false, non-causal associations due to chance, bias or confounding) Observed association between a disease and suspected factor may not be real . This is due to selection bias Eg: Increased water intake and crime rate in summer. The ringing of alarm clocks and rising of the sun. Cock’s crow causes sun to rise. 22

23. Ex : Neonatal mortality was observed to be more in the newborns born in a hospital than those born at home. This is likely to lead to a conclusion that home delivery is better for the health of newborn. However, this conclusion was not drawn in the study because the proportion of “high risk” deliveries was found to be higher in the hospital than in home. 23

24. Indirect Association It is a statistical association between a characteristic of interest and a disease due to the presence of another factor i.e. common factor (confounding variable). So the association is due to the presence of another factor which is common to both, known as CONFOUNDING factor. Ex : 1.Rahul is a friend with Suma, and Suma is Shoba’s friend, so Shoba is Rahul ’s friend too but indirectly. The common friend is Suma. 24

25. Altitude and endemic goiter confounding factor is iodine deficiency. Glucose and CHD ,confounding factor is cigarette smoking(it increase the of cups of coffee and amount of sugar u consume) 25

26. Direct Association The association between the two attributes is not through the third attributes. When the disease is present, the factor must also be present. 26

27. Direct (Causal ) association: One –to- one causal association Multifactorial causation Sufficient & necessary cause Web of causation (Interaction) 27

28. One-to-one Casual Relationship The variables are stated to be casual related (AB) if a change in A is followed by a change in B. When the disease is present, the factor must also be present. A single factor (cause) may lead to more than one outcome. But its not always that simple , as some causes can cause more than 1 disease like streptococci 28 Hemolytic Streptococci Streptococcal tonsillitis Scarlet fever Erysipelas

29. Multifactorial causation Multiple factor leads to the disease. Common in non-communicable diseases Alternative causal factors each acting independently . Ex : In lung cancer more than one factor (e.g. air pollution, smoking, heredity) can produce the disease independently. Either the causes are acting Independently OR Cumulatively 29

30. 30 Smoking   Air pollution Reaction at cellular level Lung cancer   Exposure to asbestos Smoking + Air pollution Reaction at cellular level Lung cancer + Exposure to asbestos Independently Cumulatively

31. CAUSATION 31

32. WHAT IS CAUSE The word cause is the one in general usage in connection with matters considered in this study, and it is capable of conveying the notion of a significant, effectual relationship between an agent and an associated disorder or disease in the host.” 1964 Surgeon General Report 32

33. General Models of Causation The most widely applied models are: The epidemiological triad ( triangle) , The web The wheel and The sufficient cause and component causes models ( Rothman’s component causes model ) 33

34. Epidemiological triad 34 Agent Factors Physical Agents Chemical Agents Biological Agents Nutritional agents Host Factors Socio-demographic Factors Psycho-social Factors Intrinsic Characteristics Environmental Factors Physical Environment Biological Environment Social Environment

35. Web of Causation 35 DISEASE behaviour Unknown factors genes phenotype workplace microbes environment

36. 36 Web of Causation - CHD CHD Unknown factors gender inflammation medications lipids physical activity blood pressure stress

37. 37 Social Environment Genetic Core Biological Environment Host (human) Physical Environment Wheel of Causation

38. Sufficient & Necessary Cause NECESSARY cause - causal factor whose presence is required for the occurrence of the effect. If disease does not develop without the factor being present, then we term the causative factor “ necessary ”. Ex: Agent in Malaria: Plasmodium falciparum parasite is necessary factor- always present. SUFFICIENT cause - “minimum set of conditions, factors or events needed to produce a given outcome. Usually there’s no sufficient factor “rare”. The factors or conditions that form a sufficient cause are called component causes. Necessary causes + Component causes = Sufficient cause 38

39. Rothman’s Component Causes and Causal Pies Model Rothman's model has emphasised that the causes of disease comprise a collection of factors. These factors represent pieces of a pie, the whole pie ( combinations of factors) are the sufficient causes for a disease. It shows that a disease may have more that one sufficient cause, with each sufficient cause being composed of several factors 39

40. The factors represented by the pieces of the pie in this model are called component causes. Each single component cause is rarely a sufficient cause by itself, But may be necessary cause. Control of the disease could be achieved by removing one of the components in each "pie" and if there were a factor common to all "pies“ (necessary cause) the disease would be eliminated by removing that alone. 40 A U B C N

41. 41 A U B C N Known components (causes) – A, B, C Unknown component (cause) - U N – Necessary cause Known components causes + Unknown component cause = Sufficient cause + Necessary cause

42. 42 Causes of tuberculosis Infection Tubercu-losis Susceptible host

43. 43 TYPES OF CAUSAL RELATIONSHIPS

44. If a relationship is causal, four types of causal relationships are possible: (1) Necessary And Sufficient (2) Necessary, But Not Sufficient (3) Sufficient, But Not Necessary (4) Neither Sufficient Nor Necessary 44

45. Necessary and Sufficient A factor is both necessary and sufficient for producing the disease. Without that factor, the disease never develops and in the presence of that factor, the disease always develops Types of causal relationships I: Each factor is both necessary and sufficient 45 FACTOR A DISEASE

46. Necessary, But Not Sufficient Each factor is necessary, but not, in itself, sufficient to cause the disease . Thus, multiple factors are required, often in a specific temporal sequence. Ex: Carcinogenesis is considered to be a multistage process involving both initiation and promotion. A promoter must act after an initiator has acted. Action of an initiator or a promoter alone will not produce a cancer 46

47. Types of causal relationships: Each factor is necessary, but not sufficient 47

48. Sufficient But Not Necessary The factor alone can produce the disease, but so can other factors that are acting alone Either radiation or benzene exposure can each produce leukemia without the presence of the other. Even in this situation, however, cancer does not develop in everyone who has experienced radiation or benzene exposure, so although both factors are not needed, other cofactors probably are. Thus, the criterion of sufficient is rarely met by a single factor. 48

49. Each factor is sufficient, but not necessary 49

50. Neither Sufficient Nor Necessary A factor by itself, is neither sufficient nor necessary to produce disease This is a more complex model, which probably most accurately represents the causal relationships that operate in most chronic diseases. Types of causal relationships: IV. Each factor is neither sufficient nor necessary 50

51. When we can say that this association is likely to be causation?? We have certain criteria that should be present: Temporal association Strength of association Specificity of association Consistency of association Biological plausibility Coherence of association 51

52. 52

53. Guidelines for Judging Whether an Association Is Causal (Leon Gordis ) 1.    Temporal relationship    2.    Strength of the association    3.    Dose-response relationship    4.    Replication of the findings    5.    Biologic plausibility    6.    Consideration of alternate explanations    7.    Cessation of exposure    8.    Consistency with other knowledge    9.    Specificity of the association 53

54. Temporal association The causal attribute must precede the disease or unfavorable outcome. Exposure to the factor must have occurred before the disease developed. Length of interval between exposure and disease very important . Its more obvious in acute disease more than in chronic disease 54

55. Temporal relationship (Relationship with time) Cause must precede the effect. Drinking contaminated water occurrence of diarrhea However in many chronic cases, be cause of insidious onset and ignorance of precise induction period, it become hard to establish a temporal sequence as which comes first -the suspected agent or disease. 55

56. Strength Of The Association Relationship between cause and outcome could be strong or weak. With increasing level of exposure to the risk factor an increase in incidence of the disease is found. Strong associations are more likely to be causal than weak. Weaker associations are more likely to be explained by undetected bias. But weaker association does not rule out causation. 56

57. Strength of association can be estimated by relative risk, attributable risk etc. Relative risks/Odds ratio greater than 2 can be considered strong 57

58. Dose-Response Relationship ( The Biological gradient ) As the dose of exposure increases, the risk of disease also increases If a dose-response relationship is present, it is strong evidence for a causal relationship. However, the absence of a dose-response relationship does not necessarily rule out a causal relationship. In some cases in which a threshold may exist, no disease may develop up to a certain level of exposure (a threshold); above this level, disease may develop 58

59. Death rates from lung cancer (per 1000) by number of cigarettes smoked, British male doctors, 1951 –1961 59

60. Biologic Plausibility Of The Association The association must be consistent with the other knowledge ( viz mechanism of action, evidence from animal experiments etc ). Sometimes the lack of plausibility may simply be due to the lack of sufficient knowledge about the pathogenesis of a disease. It is too often not based on logic or data but only on prior beliefs. It is difficult to demonstrate where the confounder itself exhibits a biological gradient in relation to the outcome. 60

61. Consideration of Alternate Explanations Interprets an observed association in regard to whether a relationship is causal or is the result of confounding. In judging whether a reported association is causal, the extent to which the investigators have taken other possible explanations into account and the extent to which they have ruled out such explanations are important considerations. 61

62. Cessation of Exposure If a factor is a cause of a disease, we would expect the risk of the disease to decline when exposure to the factor is reduced or eliminated 62

63. Consistency Of The Association Consistency is the occurrence of the association at some other time and place repeatedly unless there is a clear reason to expect different results. If a relationship is causal, the findings should be consistent with other data. Lack of consistency however does not rule out a causal association. Repeated observation of an association in different populations under different circumstances. 63

64. Specificity Of The Association The weakest of the criteria. (should probably be eliminated) Specific exposure is associated with only one disease. Specificity implies a one to one relationship between the cause and effect. It’s the most difficult to occur for 2 reasons: Single cause or factor can give rise to more than 1 disease Most diseases are due to multiple factors. Ex : Smoking is associated with many diseases. Not everyone who smokes develops cancer Not every one who develop cancer has smoke 64

65. Analogy (Similarity, reasoning from parallel cases) Provides a source of more elaborate hypotheses about the associations under study. Absence of such analogies only reflects lack of imagination or experience , not falsity of the hypothesis. Ex : Known effect of drug Thalidomide & Rubella in pregnancy Accepting slighter but similar evidence with another drug or another viral disease 65

66. Coherence of the association and judging the evidence Based on available evidence or should be coherence with known facts that are thought to be relevant: uncertainty always remains. Correct temporal relationship is essential; then greatest weight may be given to plausibility, consistency and the dose–response relationship. The likelihood of a causal association is heightened when many different types of evidence lead to the same conclusion. 66

67. Deriving causal inferences: example Assessment of the Evidence Suggesting Helicobacter pylori Ulcers as a Causative Agent of Duodenal    1.    Temporal relationship .    •    Helicobacter pylori is clearly linked to chronic gastritis. About 11% of chronic gastritis patients will go on to have duodenal ulcers over a 10-year period.    2.    Strength of the relationship .    •    Helicobacter pylori is found in at least 90% of patients with duodenal ulcer. 67

68. 3. Dose-response relationship .    •    Density of Helicobacter pylori per square millimeter of gastric mucosa is higher in patients with duodenal ulcer than in patients without duodenal ulcer 4.    Replication of the findings .(consistency)    •    Many of the observations regarding Helicobacter pylori have been replicated repeatedly 5.    Consideration of alternate explanations .    •    Data suggest that smoking can increase the risk of duodenal ulcer in Helicobacter pylori -infected patients but is not a risk factor in patients in whom Helicobacter pylori has been eradicated 68

69. 6.    Biologic plausibility .    •   Originally it was difficult to envision a bacterium that infects the stomach antrum causing ulcers in the duodenum, but is now recognized that Helicobacter pylori has binding sites on antral cells and can follow these cells into the duodenum.    •    Helicobacter pylori also induces mediators of inflammation.    •   Helicobacter pylori -infected mucosa is weakened and is susceptible to the damaging effects of acid.    7.    Cessation of exposure .    •   Eradication of Helicobacter pylori heals duodenal ulcers at the same rate as histamine receptor antagonists.    •   Long-term ulcer recurrence rates were zero after Helicobacter pylori was eradicated using triple-antimicrobial therapy,. 69

70. 8.    Specificity of the association .    •   Prevalence of Helicobacter pylori in patients with duodenal ulcers is 90% to 100%. 9.    Consistency with other knowledge .    •  Prevalence of Helicobacter pylori infection is the same in men as in women. The incidence of duodenal ulcer, which in earlier years was believed to be higher in men than in women, has been equal in recent years.    •   The prevalence of ulcer disease is believed to have peaked in the latter part of the 19th century, and the prevalence of Helicobacter pylori may have been much higher at that time because of poor living conditions. 70

71. Modified Guidelines for Evaluating the Evidence of a Causal Relationship. ( In each category, studies are listed in descending priority order .) 1990 1.    Major criteria    a.    Temporal relationship : An intervention can be considered evidence of a reduction in risk of disease or abnormality only if the intervention was applied before the time the disease or abnormality would have developed.    b.    Biological plausibility : A biologically plausible mechanism should be able to explain why such a relationship would be expected to occur.    71

72. c.    Consistency : Single studies are rarely definitive. Study findings that are replicated in different populations and by different investigators carry more weight than those that are not. If the findings of studies are inconsistent, the inconsistency must be explained.    d.    Alternative explanations (confounding ): The extent to which alternative explanations have been explored is an important criterion in judging causality 72

73. 2.    Other considerations    a.    Dose-response relationship: If a factor is the cause of a disease, usually the greater the exposure to the factor, the greater the risk of the disease. Such a dose-response relationship may not always be seen because many important biologic relationships are dichotomous, and reach a threshold level for observed effects.    b.    Strength of the association: Usually measured by the extent to which the relative risk or odds depart from unity. c.    Cessation effects: If an intervention has a beneficial effect, then the benefit should cease when it is removed from a population. 73

74. Modern concepts in causation Counterfactual Model Causal diagram 74

75. Counterfactual model (Potential outcome model) When we are interested to measure effect of a particular cause, we measure effect in a population who are exposed. We calculate risk ratios & risk differences based on this model The difference of the two effect measures is the effect due the cause we are interested in. 75

76. Causal Diagram Confounding is complex phenomenon. Useful for analysis of confounders Conceptual definition of variable involved Directionality of causal association Need some level of understanding (Knowledge & hypothetical) – relation between risk factor, confounders & outcome. Directed Acyclic Graph (DAG) 76

77. ANALYTICAL METHOD – ASSOCIATION MEASURES 77

78. Analytical Methods Measures of association /strength of association Testing hypothesis of association Controlling confounders 78

79. Measures of association Ratio measures Measures of association in which relative differences between groups being compared Difference measures Difference measures are measures of association in which absolute differences between groups being compared . 79

80. Absolute differences:( difference measures ) Main goal is often an absolute reduction in the risk of an undesirable outcome. When outcome of interest is continuous, the assessment of mean absolute differences between exposed and unexposed individuals may be an appropriate method for the determination of association. Relative differences: ( ratio measures ) Can be assessed for discrete outcomes. To assess causal associations 80

81. 81

82. Relative risk If an association exist, then how strong is it? What is the ratio of the risk of disease in exposed individuals to the risk of disease in unexposed individual? Incidence among exposed Relative risk = Incidence among unexposed It is direct measure of the strength of association. 82

83. 83

84. Relative risk of developing the disease is expressed as the ratio of the risk(incidence) in exposed individuals (q+) to that in unexposed individual(q-) Total exposed = a+b Total unexposed = c+d 84

85. Incidence among exposed Relative risk = Incidence among unexposed a/ a+b RR = q+/q- = c/ c+d 85

86. 86 Odds ratio in a cohort study Odds that an exposed person develop disease = a/b Odds that an unexposed person develop disease = c/d Odds ratio = (a/b ) / (c/d) = ad/ bc Develop disease Do not develop disease Exposed a b Unexposed c d What are the odds that the disease will develop in an exposed person?

87. 87 Relationship between OR and RR OR is a valid measure of association in its own right and it is often used as an approximation of the relative risk’. Use of OR as an estimate of the relative risk biases it in a direction opposite to the null hypothesis, i.e. it tends to exaggerate the magnitude of the association.

88. 88 ATTRIBUTABLE RISK (AR) AR is defined as the amount of proportion of disease incidence (or disease risk) that can be attributed to a specific exposure. Based on the absolute difference between two risk estimates. Used to imply a cause-effect relationship and should be interpreted as a true etiologic fraction only when there is a reasonable certainty of a causal connection between exposure and outcome.

89. 89 AR in exposed individuals It is merely a difference between the risk estimates of different exposure levels and a reference exposure level. If q + = risk in exposed individual. q - = risk in unexposed individual. AR exp = q + - q - It measures the excess risk for a given exposure category associated with the exposure

90. 90 Percent AR exposure When AR is expressed as a percentage Interpretation: The percentage of the total risk in the exposed attributable to the exposure.

91. 91 POPULATION ATTRIBUTABLE RISK What proportion of the disease incidence can be attributed to a specific exposure in a total population . To know the PAR , we need to know incidence in total population =a incidence in unexposed group(background risk)=b PAR= a-b ÷ a

92. 92 Various correlation tests Pearsson’s product-moment correlation Spearmans rank order correlation Kendall correlation Point biserial correlation Tetrachoric correlation Phi correlation

93. 93 Types of correlation Based on linearity of correlation

94. 94 Based on direction of correlation Positive correlation: As X increases ,Y also increases, ex: As height increases, so does weight. Negative correlation: As X increases ,Y decreases. ex: As time of watching TV increases , grade scores decreases.

95. Perfect positive Moderately positive Zero correlation Moderately negative Perfectly Negative 95 Based on degree of correlation

96. 96

97. 97

98. 98

99. REGRESSION It can also be used in measuring association. They are the measure of the mean changes to be expected in the dependent variable for a unit change in the value of the independent variable. When more than 1 independent variable is associated with the dependent variable, multiple regression analysis will indicate how much of the variation observed in the dependent variable can be accounted for, by one or a combination of independent variables. 99

100. PROBLEMS IN ESTABLISHING CAUSALITY The existence of correlation/ association does not necessarily imply causation. Concept of single cause  concept of multiple causation Koch’s postulates cannot be used for non-infectious diseases. The period between exposure to a factor and appearance of clinical diseases is long in non-infectious diseases. Specificity established in one disease does not apply on others. 100

101. Confounders associated with disease tend to distort relationship with the suspected factors. Systematics errors/ bias can produce spurious association. No statistical method can differentiate between causal and non-causal. Because of these many uncertainties, the terms : Causal inference, causal possibility, or likelihood are preferred to causal conclusion. This helps in formulating policy rather than waiting for the unequivocal proof ( Unattainable in several disease conditions) 101

102. 102 Results from epidemiological studies are often used as inputs for policy and judicial decisions. It is thus important for public health and policy makers to understand the fundamentals of causal inference. Association does not imply causation. Apart from outbreak investigations, no single study is capable of establishing a causal relation or fully informing either individual or policy decisions. Those decisions should be based on a carefull consideration of the entire relevant scientific and policy literature Conclusion

103. 103 [1] Park K. Textbook of Preventive and Social Medicine. 23 rd ed. [2] Gordis , Leon Epidemiology / Leon Gordis .—5th ed. [3]Roger Detels et al. Oxford Text Book of Public Health. 5 th ed. New york (U.S.A): Oxford University Press; 201 WHO research methodology. Second edition. AFMC WHO – Text book of Public Health and Community Medicine – Rajvir Balwar – 1 st edition Soben peters – Text book of Community Dentistry – 5 th edi Raj Bhopal : Cause and effect: the epidemiological approach : Google book source References

104. 104 THANK YOU