EPI 390 1st Edition Lecture 1 Outline of Current Lecture I. VocabularyII. Understanding BiasIII. Inductive vs. Deductive ReasoningIV. Defining CausalityV. Designing Epidemiological ExperimentsCurrent LectureI. Vocabularya. Relative Risk – The ratio of probability of an event occurring between an exposedgroup and a non-exposed group.i. Relative Risk as a negative number: Vaccine exposure.b. Odds Ratio - the ratio of the odds of an event occurring in one group to the odds of it occurring in another group, or to a data-based estimate of that ratio; estimates the probability of disease given exposure to a specific factor by measuring the probability of exposure given the presence of disease. i. Odds ratio compared to general population of 3 = 3 times more likely to have outcome than unexposed publicii. Odds Ratio for lung cancer and smoking range from 5-40c. Covariance – the correlation between two or more random variants.II. Understanding Biasa. Understanding how bias fits into what we do, and understanding world w/o bias.i. Humans have a propensity to read into results based on already established beliefs/opinionsii. Studying the cause of an epidemiological issue requires an experiment that limits bias (see bullet V.)b. Association doesn’t necessarily mean relationIII. Inductive vs. Deductive Reasoninga. Inductive (aka instinctive) reasoning – A conclusion is drawn from specific instances/examplesThese notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.i. Premises of the conclusion are based off of facts and observations, though it is possible for facts to be correct while conclusion is false.ii. Inductive reasoning is mostly descriptive. It involves stating ideas and generating hypotheses or discovering relationships, only as the very beginning. 1. Very simple. iii. Example: There are more females than males in this course. You can count the number of females v. males, but the result is merely descriptive. Need to continue on to ‘why?’, ‘what does it mean for the population?’.b. Deductive (aka ingenious) Reasoning – A conclusion is drawn from general principles or premises.i. The premise is a previously made statement from which another conclusion is inferred.ii. Conclusions from deductive reasoning are certain, unlike the uncertainty in inductive reasoning, provided the premises are true.IV. Defining Causalitya. Temporal Relation – Cause must precede effect (this can be difficult to determinein some cases)i. Just because one event happens first, doesn’t mean that it was the cause of the effect (post hoc conclusions)b. Plausibility – the conclusion coincides with existing knowledge/evidencei. A lack of plausibility may simply be due to a lack of knowledge. You can’t know what you don’t know.c. Consistency – Multiple studies give the same resulti. If 100 other tests get Z and you get X, you might be on to something new, but chances are you: didn’t have right variables, made mistake in stats, etc.ii. When comparing to previous studies, emphasis should be placed on studies with the most sound design.d. Strength – a logical association with cause and effect. i. Looks at relative riske. Dose-Response Relationship – Varying the amount of exposure from the event results in changing magnitudes of the effect.f. Reversibility – Removal of the possible cause results in a reduction of disease risk.V. Designing Epidemiological Experimentsa. Main point of epidemiology is to gather info on public to give to public health officials so that they can change something.b. Is the evidence based on a strong enough study design?i. Sample size – a large sample size makes results more statistically significant by including a wide variety of individuals/variants/etc.ii. Sample Composition – reducing bias by having a diverse sample population, therefor controlling for multiple variants.iii. Reference Group – the reference group controls for all but one variant.iv. Anonymity – reduces bias and protects the individuals studied.g. Descriptive epidemiology looks at frequencies (count data)i. Statistics are necessary for quantifying continuous datah. Latent variables – variables that you cannot measure directly; use substitute measures (get at them through other means)i. An experimenter can put together questionnaire that can kinda guess at issues (i.e. were you depressed this morning?) but there is no direct measure like in metric measuresii. Outside influence: feeling anxious on normal class day vs. exam dayiii. Stress, addiction (behavioural manifestations), OCDi. Covariance: other potential variables in outcome (i.e. cantaloupe listeria: what else did they buy/eat? Was it really the cantaloupe or one of these other things?)j. Contingency tables: outcomes for exposure (you were exposed, you weren’t) x outcomes for disease (you have it, you