Concept Of Causation In Epidemiology PDF

Objectives Concept of Causation in Epidemiology Observed associations for assigning...

Objectives Concept of Causation in Epidemiology Observed associations for assigning or explaining causal or non-causal relationships are prone to misapplication as it relates to diseases causation. The observed association At the end of this lecture students should be able to: could be the result of random error, such as 1. define the terms such as causal factor chance, or a systematic error in research design 2. discuss some features of causal factors or analysis leading to bias or even a 3. describe the process of establishing causation confounding factor, in the case where the disease and the exposure are associated with 4. discuss the Bradford Hill’s criteria for helping another factor. to establish causation For example, it is said that sedentary lifestyle can result in cardiovascular disease. However, it is also true that obesity and bad eating habits, resulting in high cholesterol levels, can also cause cardiovascular problems. It can be a challenge therefore to ascertain conclusively, the causative agent for cardiovascular disease in someone who has all three exposures. Determinants of diseases may also be influenced by genetic or environment. Consequently, absence of an association does not necessarily mean absence of causation (Fletcher and Fletcher, 2005). Establishing a cause-effect relationship is important for prevention but it is of equal importance to correct diagnosis and treatment. There are differences in judgment regarding the concept of causation (Beaglehole and Kjellstrom, 2008, Holfer, 2005). How can one determine if unemployment is the causal factor for malnutrition or cardiovascular disease? Firstly, the factors that are necessary for malnutrition must be present in sufficient quantities and the status of unemployment must be clearly defined. Proving that these factors are associated and, more so, causal, requires rigid investigation and explicit details. Sir Austin Bradford Hill (1897-1981) developed some criteria which are to guide this process of establishing causation in epidemiological studies. It must, however, be explicitly noted that no one criterion, except for temporal relationship, is used by itself to establish causation, with clear understanding of associations and causation. The words of Hill “None of these nine viewpoints can bring indisputable evidence for or against a cause and effect hypothesis …. What they can do, with greater or less strength, is to help answer the fundamental question—Is there any other way of explaining the set of facts before us? Is there any other answer equally, or more, likely than cause and effect?” (Doll, 1991). Along with statistical calculations, such as relative risk and odds, and rigorous research designs the likes of cohort studies and case control studies, associations and causation for various diseases are established. A causal factor is any event, behaviour, condition, deficiency, omission or characteristic that if changed, by means of elimination, alteration or avoidance would increase the likelihood of disease or probably prevent fatality (Jeneick, 2010). There are two major categories of causal factors: direct factors which lead directly to changes in the cells of the body and indirect factors, which increase the likelihood of pathogenic changes. Single event: for example one contact with the HIV virus. Complex event: where there are interactions with several factors otherwise termed as a causal web or web of causation. An example of complex event involves  chronic non-communicable diseases such as Diabetes Mellitus or Hypertension. Obesity in those over forty year old, a sedentary life, hereditary factors and a high calorie diet, are all causal factors for cardiovascular diseases. Some causal factors may be described as necessary factors in that they must be present for the disease to occur. An example of this would be the tubercle bacillus in Tuberculosis. This is a complex interaction, where the bacteria must be present and the immune status compromised, in a particular environment. The cause is termed necessary if an outcome cannot develop in its absence. A necessary cause represents a condition that must be present for the effect to follow and is sufficient, when it inevitably produces an outcome. A sufficient cause represents a condition that will guarantee the effect. Being female is a necessary condition of being pregnant. Being female is not a sufficient condition, since you can be female without being pregnant. Being convicted of a crime is a sufficient cause of being judged guilty; it is not a necessary cause, however, since you could be judged guilty as a result of confessing to the crime. Having sexual intercourse, for example, would be called a necessary cause of being pregnant; yet it is observed that it is not absolute, since you could get pregnant via artificial insemination. However, since the majority of pregnancies result from intercourse, it makes sense to call it a necessary cause. It is clearly not a sufficient cause, since most of the time, intercourse does not result in pregnancy (Gordis, 2006, Beaghole and Kelljstrom, 2008). Necessary and sufficient: Without the factor, disease can never develop. On the other hand, with the factor, disease always develops (Note: This situation rarely occurs). It is rare because in most infectious diseases a number of people are exposed; some of whom will manifest the disease while others will not. In members of households with Tuberculosis – if the exposure is the same - the disease will or will not develop due to immune status of individuals, genetic susceptibility or any other characteristics that can influence the development of the disease. Necessary but not sufficient: The factor in and of itself is not enough to cause disease. Multiple factors are required, usually in a specific temporal sequence. Carcinogenesis, a multistage process, is a principal example. For cancer to result a promoter must act after an initiator has acted. Also for TB, if sufficient level of the tubercule bacillus (a necessary factor) is not present, the disease will not develop. Sufficient but not necessary: The factor alone can cause disease, but so can other factors in its absence. Benzene or radiation can cause leukemia, without the presence of the other. So the factor alone can produce the disease, but so can other factors that are acting alone. Either radiation exposure or benzene exposure produces leukemia without the presence of the other. Since cancer does not develop in all who are exposed, the criterion of sufficiency is rarely met as a single factor. Neither sufficient nor necessary The factor cannot cause disease on its own, nor is it the only factor that can cause that disease. This is the probable model for chronic disease relationships. There are seldom absolute cases of either necessary or sufficient cause and rarely, if ever, have absolute cases of both causes. Did I understand? FOR EACH OF THE FOLLOWING FOLLOWING RISK FACTORS AND AND HEALTH OUTCOMES, IDENTIFY ANDAND INDICATE ON THE LI LINE WHETHE THEY ARE NECESSARY CAUSES, SUFFICIENT SUFFICIENT CAUSES, OR NECESSARY BUT NOT SUFFICIENT SUFFICIENT. FFICIENT. RISK FACTOR HEALTH OUTCOME ________________ HYPERTENSION HYPERTENSION STROKE ________________ TREPONEMA TREPONEMA PALLIDUM SYPHILLIS ________________ TYPE TYPE A PERSONALITY HYPERTENSION ________________ SKIN SKIN CONTACT WITH A STRONG STRONG ACID BURN Establishing the process of causation Stages in establishing the process of Because of the level of ambiguity in establishing causation, several criteria causation: were developed to help in establishing causation, and there are stages in the process. Statistical association between exposure to the causal factor Stage 1 and the disease is the first stage. The occurrence of disease causation is Establish Statistical Association derived from descriptive statistics - median, mean and mode, as well as (SA) from the second state inferential statistics. Because it is not ethically Stage 2 Establish Causal Interference possible to undertake all types of experimental research, some inferences are needed to establish association and further causation. Evidence of causation is important. For example, in infectious disease, what evidence do we have of the causative agents? Henle in 1840 proposed postulates for causation and Koch in 1880s expanded on the following: The organism is always found with the disease. The organism is not found with any other disease The organism, isolated from one who has the disease, and cultured through several generations, produces the disease (in experimental animals). In determining whether an association is causal and to make judgment on causation depends on the type and extent of the evidence. (Fletcher and Fletcher, 2005).  WEB of CAUSATION www.pitt.edu/~super7/8011-9001/8351.ppt [Accessed 25/9/2012] The diagram provided above, depicts the complexity of disease. There are several postulations as to how disease occur: Miasmatic theory, as suggested by Farr, held the belief that diseases are transmitted by a cloud (Gordis, 2004;p:11); Henle-Koch believes diseases occur as a result of a germ (Fletcher and Fletcher,2005). The diagram, however, illustrates the fact that cause and effects of disease are much more subtle and multifaceted than such beliefs. In establishing causation, the application of some critical criteria is necessary, as proposed by Bradford Hill. Bradford Hill’s Causation 1. Strength of association The stronger the association, the more likely it is that the relation is causal. The strength is measured by the relative risk or the odds ratio. The larger risk ratio, the stronger will be the evidence for causation. A small association does not mean that there is not a causal effect, although the larger the association, the more likely that it is a causal relationship. Chimney sweeps died of scrotal cancer at rates 400 times that of the normal population. Strong association between scrotal cancer and chimney sweeping is good evidence in favour of causality from an environmental exposure. Small effects in a population, however, can still be considered strong associations. For example, Snow’s evaluation of the 1855 cholera outbreak, showed death rates from the contaminated water at 17/10,000 vs. 5/10,000 in the general population. This was not a huge increase in mortality, but it showed strong association nonetheless. 2. Consistency with other studies It is important to show similarity between findings in studies using diverse methods of study, in different populations and under a variety of circumstances. The greater the number of consistent studies, the stronger the evidence is. The studies must be free from bias. Consistent findings observed by different persons, in different places and with different samples, strengthen the likelihood of an effect.
3. Specificity of association This criterion states that the causal factor should lead to one disease only; the disease should result from the single cause. This criterion is weak. Specificity however, provides additional support for causal relationships. Causation  is likely if a very specific population, at a specific site, has a specific disease, with no other likely explanation. The more specific an association between a factor and an effect is, the bigger the probability of a causal relationship. From the perspective of opportunistic infections, with no knowledge of viral pathophysiology, HIV is hardly a specific cause of disease. Given the ability to measure HIV viral load and an understanding of the consequences of HIV depletion of CD4 cells, HIV has high specificity for causing AIDS. An association is specific when a certain exposure is associated with only one disease. Hence cigarettes smoking would not qualify for specific association. (Beaghole and Kelljstrom, 2008).
4. Temporal relationship This requires that the exposure to the causal factor precedes the onset of disease. This is not always easy to ascertain, particularly in the case of control studies – cohort study. The effect has to occur after the cause (and if there is an expected delay between the cause and expected effect, then the effect must occur after that delay). A prospective study design is most suitable to determine temporal relationships. For example asbestos has been linked to lung cancer, but the latent period is approximately 15-20 years. If lung cancer develops after only three years of exposure to asbestos, it is safe to conclude that the lung cancer was not the result of this exposure? (Beaghole and Kelljstrom, 2008).
5. Biological plausibility This refers to the fact that causation should be in keeping with current biological knowledge, including pathophysiology of disease. This is sometimes difficult to prove as some epidemiological observations may  precede biological knowledge. Greater exposure should generally lead to greater incidence of the effect. However, in some cases, the mere presence of the factor can trigger the effect. In other cases, an inverse proportion is observed: greater exposure leads to lower incidence. Coherence with other body of knowledge usually based on preceding epidemiologic observations, e.g. John Snow and cholera and high oxygen concentration and the development of retrolental fibroplasias. This condition was discovered before any biological knowledge was available to support the findings. When biological knowledge is not available to support the claim for causation it means that the studies must be more rigorous and be replicated where possible.
6. Dose-response relationship This criterion holds that an increase in level, intensity, duration or total level of exposure to an agent leads to progressive increases in risk. The risk for lung cancer increases with significant increases in the number of cigarettes smoked. If an association is causal, then we would expect that the greater the exposure (dose), then the greater the effect on the outcome (response). For example, if a heavy lifetime smoker had a lower risk of lung cancer than a light smoker, or a person who smoked for only a few years, we would have to question whether smoking could be causing lung cancer. However, most cases show that heavy lifetime smokers are at greater risk for lung cancer.
7. Consistency with other available evidence This criterion holds that evidence concerning the natural history, biology and epidemiology of the disease should be consistent. They should form a cohesive whole. The proposed causal relationship should not conflict or contradict information from experimental, laboratory, clinical, pathological and epidemiological sources of knowledge.
8. Experimentation  This requires experimental epidemiological studies. They may be natural experiments where human groups are observed; eg. John Snow… Clinical trials and community trials are strong proofs of causation if the researcher is able to control the implementation of the study in order to reduce the effects of confounding factors. However, epidemiological experimentation for causation is often impractical and unethical. In vitro experiments using animals to support causation are often carried out.
9. Analogy
Analogy implies a similarity between findings from different studies. Example of this in epidemiology is that one pharmaceutical drug such as thalidomide causes severe birth defects, so might others. This criterion does not provide hard and fast evidence of cause and effects. In fact no single criterion can be used as a necessary condition or indispensable need. Instead a combination of facts, judgment and experimental support is required. (Gorman, 2012; Beaghole and Kelljstrom, 2006) Conclusion
Temporal relation - Does the cause precede the effect? (essential)
Plausibility - Is the association consistent with other knowledge? (mechanism of action; evidence from experimental animals)
Consistency - Have similar results been shown in other studies?
Strength - What is the strength of the association between the cause and the effect? (relative risk) 
Dose–response relationship - Is increased exposure to the possible cause associated with increased effect?
Determining whether an association is causal is important to the interpretation of research evidence.
Association should be independent of confounding factors. References:
Beaglehole Bonita, R and Kelljstrom (2006) Causation in Epidemiology‘Basic Epidemiology’ 2nd ed., ch: 5; pp:83- 94
Doll, R. (1991). Sir Austin Bradford Hill and the progress of medical science. British Medical Journal, 305, 1521- 1526. http://www.bmj.com/content/305/6868/1521 [Accessed 6 January 2014]
Höfler M. (2005). The Bradford Hill considerations on causality: a counterfactual perspective. Emerging Themes in Epidemiology [Online] Available from: http://www.ncbi.nlm.nih.gov/pubmed/16269083 [Accessed 6 January 2014]
Gordis, L. (2004). Epidemiology, ed., 3rd Elsevier Inc.
Gorman, S (2012). The Pump Handle What “causes” disease?: Association vs. Causation and the Hill Criteria. Public Health Series[Online] Available from: http://scienceblogs.com/thepumphandle/2012/11/19/what-causes- disease-association-vs-causation-and-the-hill-criteria/ [Accessed 6 January 2014]
Fletcher R., and Fletcher, S. (2005). “Cause” in Clinical Epidemiology, ch 11:pp: 188-199 ed., 4th Lippincott Williams and Wilkins
Jeneick, M. (2010). Medical Error and Harm: Understanding, Prevention, and Control [Online] Available from: http://books.google.com.jm/books?id=fJw- PHjveogC&pg=PA84&lpg=PA84&dq=Causal+Factors+are+any+behavior,+omission,+or+deficiency+that+if+corrected,+elimin ated,+or+avoided+probably+would+have+prevented+the+fatality&source=bl&ots=DAEzewZf4S&sig=- Jj0Jws4Jg0Auu8YvpSaFkmaJtI&hl=en&sa=X&ei=w9fKUqrnGIWdkQfm4oGoCA&ved=0CDQQ6AEwAg#v=onepage&q=Caus al%20Factors%20are%20any%20behavior%2C%20omission%2C%20or%20deficiency%20that%20if%20corrected%2C%20 eliminated%2C%20or%20avoided%20probably%20would%20have%20prevented%20the%20fatality&f=false [Accessed 6 January 2014]