Epidemiology assignment due on 2/21/15, What is a cause? Please answer questions

Excelsior College PBH 321

Page 1

HOW DO WE KNOW WHEN SOMETHING MEETS THE DEFINITION OF A CAUSE?

Remember that epidemiological studies determine an association between exposure and disease. Association is not equal to causation. Consider the following statement: If the rooster crows at the break of dawn, then the rooster caused the sun to rise. Yet there is an association between the rooster crowing at dawn and the sun rising. Steps in Evaluating Epidemiological Associations When thinking about whether an observed association between exposure and disease is causal, we have to consider whether the relationship is either due to chance (due to random error) and if it is valid (in other words, unbiased or free of systematic error). There are statistical approaches to determine if the association is what we call “statistically significant”, or unlikely to be due to chance alone. As you have learned about in Module 5, bias and confounding can introduce systematic error. Thus, when we observe a relationship between exposure and outcome, we must consider if there are alternative explanations for the observed association. If the alternative explanation is not bias and confounding - because you think you’ve controlled for them well- then the association is valid. Finally, we have to think about whether the observed association is causal. Generally, we cannot visualize causal relationships directly, but instead infer their existence. We cannot “prove” causation, but can create a belief in a causal relationship by demonstrating a causal framework. Characteristics of a Cause Causes can be host-related (e.g., genetics, gender, age, etc.) or environmental factors (e.g., conditions, actions of individuals, events, natural, social or economic phenomena). There are three major characteristics, which determine if a factor is a cause.

• Time order: causes must precede the effect. They may be either proximate to each other in time or distant, but precedence is required. Can you recall a study design, which often has a particular problem establishing this characteristic?

• Direction: of effect must be asymmetrical. Cause leads to effect, but not vice versa. We can apply this characteristic for either positive (presence of a causative exposure) or negative (lack of a preventive exposure) effects.

• Association: causes and their effects must occur together. This means there must be a statistical dependence between the causal factor and the effect. In other words, the measure of association must be greater than or less than the null value (1.0).

Causes Versus Risk Factors We have spent some time throughout this course in calculating measures of association, a measure of an observed relationship between two factors. Association means that two events are statistically related, but does not necessarily mean that one event caused the other. For example, democratic candidates are more likely to win presidential elections in years when an American League team wins the World Series –there is an observed association between democratic candidate and American League wins. However, the winner of the World Series does not cause a candidate to win a presidential election. Thus, association does not necessarily equate to cause.

Excelsior College PBH 321

Page 2

To make an analogy to the courtroom, associations observed in epidemiological studies may provide circumstantial evidence implicating a risk factor as a cause of disease. However, epidemiological association alone does not provide proof beyond a reasonable doubt. Associations must be judged in light of current knowledge and evaluated according to the likelihood that they represent causal relationships. Example: Known risk factors for breast cancer. Are these direct causes of breast cancer? In fact, they serve only as proximate causes, meaning they indicate something else that may directly affect risk. Women who are unmarried are less likely to have children. Childlessness is associated with some characteristics of breast tissue (due to differences in hormonal exposures compared to women who have been through pregnancy) that may make these women more susceptible to breast cancer.

Characteristics High Risk Group for Breast

Cancer Low Risk Group for Breast

Cancer

Country of birth North America, Northern Europe

Asia, Africa

Socioeconomic Status High Low

Marital status Never married Ever married

Religion Jewish Seventh Day Adventist, Mormon

Table 15-2, Aschengrau and Seage, Essentials of Epidemiology in Public Health