Question: How helpful did you find directed acyclic graphs in understanding the strengths and limitations of observational studies?Studies to address questions of cause & effectBrief description of how it works Several different study designs are available to use to find out if an exposure causes a public health condition $($e$.$g$.,$ an outcome$)$ so we can identify an intervention. Each study design has unique key characteristics along with similarities. They also each have strengths and weaknesses. So$,$ when we begin, a feasible study design best suited to answer research questions chosen. In most cases, a study sample is recruited from a population and data collection about exposure and outcome status $($among other factors$)$ begins $($i$.$e$.,$ administration of surveys, lab testing, etc.$).$Internal validity of studies of cause and effectAs mentioned, after data collection is complete, we then want to analyze the data we collected by calculating the measure of association$($eg$.,$ odds ratio or risk ratio$)$ so we can determine whether an exposure increases or decreases the risk of the public health condition. $($e$.$g$.,$ to identify an intervention$).$Then, ultimately, we want to take that result and be able to apply it to the larger population from which we drew the sample.But there is just one problem.Be aware that the relative risk, that we calculated and interpreted, may be invalid due to confounding and$/$or selection bias. This results in a study that has poor internal validity $-$ quite simply, this means the relative risk that we calculated is inaccurate $($eg$.,$ biased$)$ and does not reflect the true effect of exposure on disease. This ultimately means you may have misidentified an intervention.Among other concepts, today we will learn about factors that affect the validity of study results as well as which study designs are more prone to poor internal validity. Observational Studies $($Case Report; Case Series; Cross$-$Sectional; Cohort;Case$-$Control; Ecologic$)$ Researcher does not have control over assignment and manipulation of exposure. Can be difficult to conduct an observational study that provides an unbiased estimate of the effect of exposure on outcome, particularly for chronic diseases. So$,$ we typically describe these studies as showing associations. Experimental Studies $($Randomized Controlled Trials $-$ RCTs$)$ Researchers have "control" over the assignment and manipulation of exposure.May provide unbiased estimate of the effect of exposure on outcome, i$.$e$.,$ demonstrate causality, so we typically describe these studies as showing causality Expensive and not always feasibleQuasi$-$Experimental Studies Naturally$-$occurring experiments Increasingly commonly usedCompetencies$1.$Apply epidemiological methods to the breadth of settings and situations in public health practiceIdentify major observational study designs $($of cause and effect$)$ Appraise threats to internal validity in observational studies of cause and effect, such as confounding, selection bias and information bias, using directed acyclic graphs$2.$ Apply concepts from relevant scientific disciplines, such as toxicology and physiology, to anticipate effects of environmental, occupational and nutritional exposures on both human health and overall planetary health.Select quantitative and qualitative data collection methods appropriate for a given public health context$4.$ Apply systems thinking tools to a public health issueFinding interventions Search systematically? Guess from observations? Clinical insights Apply a theory? Use a scientific model?Importance Early detection of unusual occurrences or presentations Preliminary source of information for potential exposures Can often be the start of something big such as the case series of pneumonia among homosexual males.Limitations Do not have comparison group Usually very small sample size Often do not systematically collect data $($rely on medical charts$).$ A cohort is a group of individuals with similar characteristics that can be assembled from:$-$ Defined population $($based on some geographic, demographic or other common characteristic,$-$ Occupational group$-$ Exposure$-$based $($based on whether or not they had the exposure of interest$)-$ Education cohort $($based on age$)$Here we start with a group of people, without the disease of interest then assess their exposure status at a "baseline visit" Then follow them over time to determine the risk of disease. Were those who were exposed were more or less likely to develop the disease of interest? Can calculate risk ratio $($and odds ratio when disease is rare$).$Directed acyclic graphs $($DAGs$)$ are incredibly helpful in understanding the strengths and limitations of observational studies, particularly in assessing the internal validity of cause$-$and$-$effect relationships. They provide a visual representation of the relationships between variables and potential confounders, helping researchers identify and control for biases like confounding and selection bias. By mapping out these relationships, researchers can better interpret study results and make more accurate conclusions about the effects of exposures on outcomes.