SERious EPI

In this episode we feature a super expert on all things related to selection bias, Dr. Chanelle Howe. There are a lot of confusing issues related to selection bias: how it’s defined, how it relates to collider stratification bias, whether it’s a threat to internal or external validity (or both!). Chanelle helps us understand many of the nuances related to selection bias and provides helpful resources for readers interested in learning more about the topic. Is a lack of exchangeability related to confounding bias or selection? How can DAGs help us decipher the difference between confounding bias and selection? Can you have selection bias in a prospective cohort study? Join us to find out the answers to all of these questions and much more! Resources: Hernán MA. Invited Commentary: Selection Bias Without Colliders. Am J Epidemiol. 2017 Jun 1;185(11):1048-1050. doi: 10.1093/aje/kwx077. PMID: 28535177; PMCID: PMC6664806. Lu H, Cole SR, Howe CJ, Westreich D. Toward a Clearer Definition of Selection Bias When Estimating Causal Effects. Epidemiology. 2022 Sep 1;33(5):699-706. doi: 10.1097/EDE.0000000000001516. Epub 2022 Jun 6. PMID: 35700187; PMCID: PMC9378569. Howe CJ, Cole SR, Chmiel JS, Muñoz A. Limitation of inverse probability-of-censoring weights in estimating survival in the presence of strong selection bias. Am J Epidemiol. 2011 Mar 1;173(5):569-77. doi: 10.1093/aje/kwq385. Epub 2011 Feb 2. PMID: 21289029; PMCID: PMC3105434.

In this episode, Matt and Hailey discuss all things selection bias. This chapter on selection bias and generalizability is the shortest of the bias chapters in the Modern Epidemiology textbook. Does that mean it’s the simplest? Listen to this episode and decide for yourself!

In this episode we have a conversation with Patrick Bradshaw about issues related to measurement error, misclassification, and information bias. We ask him to help define and clarify the differences between these concepts. We chat about dependent and differential forms of misclassification and how helpful DAGs can be for identifying these sources of bias. Patrick helps to explain the problem with the over-reliance on non-differential bias producing a bias toward the null and concerns about being “anchored to the null” in epidemiologic analyses. This episode will also serve to provide you with the most up-to-date information from Patrick on his recommendations about excellent new TV shows to stream (Wednesday on Netflix; Wandavision on Disney+). Two thumbs up.

In the season three premiere Matt and Hailey discuss Chapter 13 in Modern Epidemiology, 4th edition. For the third season of the SERious Epi podcast, we are going to continue our close-reading of the newest version of the Modern Epi textbook. This chapter is focused on measurement error and misclassification. In this episode we discuss issues related to the mis-measurement of exposure, outcome, and covariates. We also debate whether misclassification is just an analytic issue (i.e., putting people into the wrong categories) or an analytic + conceptual issue (i.e., putting people into the wrong categories and having an incorrect definition for those categories). We also talk about measurement error DAGs, why we wish more people use analytic approaches to correct for measurement error, and Matt explains the concept of email bankruptcy.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt connect with Dr. Jon Huang for a discussion on precision and study size. We wade into whether or not we should use p-values. We discuss whether the debates on p-values are real or just on Twitter and whether they should be used in observational epi or just in trials. We ask whether p-values do more harm than good in observational studies or whether the harm is really around null hypothesis significance testing. We talk about misconceptions about p-values. And Jon tells us how he’s going to win a gold medal in the Winter Olympics, despite living in a tropical climate.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt finally start talking about random error. We explore the deep philosophical (as deep as we are capable of) meaning behind randomness and whether the universe is a random (and hey, while we are at it, is there even free will) and how we think about random error. We talk about p-hacking and p-curves and anything p really. And we talk about precision and accuracy in epidemiologic research. And Hailey aces Matt’s quiz.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt connect with Dr. Maya Mathur for a discussion on confounding. We talk about different ways of thinking about confounding and we discuss how different sources of bias can come together. We talk about overadjustment bias, a topic we all feel needs more attention. We discuss e-values, and have Dr. Mathur explain their practical utility and also how complicated they are to interpret. And we discuss bias analysis for meta-analyses. Article mentioned in this episode: Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology. 2009 Jul;20(4):488-95. doi: 10.1097/EDE.0b013e3181a819a1. PMID: 19525685; PMCID: PMC2744485.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt discuss confounding and whether confounding is hogging the spotlight in epi methods and epi teaching. We debate the value of all the different terms for confounding in the world of epi and beyond and struggle to define them all. We talk about different definitions for confounding and we differentiate between confounders and confounding. We talk about the 10% change in estimate of effect approach and its limitations and we talk about different strategies for confounder control. And Hailey coins the term “DAGmatist”. We reference the paper below: VanderWeele, T.J. and Shpitser, I. (2011). A new criterion for confounder selection. Biometrics, 67:1406-1413.

In this episode of Season 2 of SERious Epidemiology, (recorded back when we were getting COVID booster shots) Hailey and Matt connect with Dr. Ellie Matthay for a discussion on Chapter 8 on case-control studies. We finally answer whether it is spelled with a – or not (and Hailey and Ellie disagree with Matt about semicolons). We discuss how cohort studies and case control studies differ and overlap. We talk about whether case control studies are more biased than cohort studies. And Hailey reveals her dreams for releasing Modern Epidemiology: the Audiobook (with possible singing).

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt get into the humble case control study. We discuss the ins and outs of this much maligned study design that has so flummoxed so many in epidemiology. We ask the hard questions about the best way sample in a case control study, whether we spend too much or not enough time on it in our teaching, whether a case control study always has to be nested within some hypothetical cohort, whether the design is inherently more biased than cohort studies (spoiler: no, but…), why some people refer to cases and controls when they are not referring to a case control study, and, if it were on a famous TV show, which character the case control study would be (and more importantly, why Hailey has never seen said TV show). Papers referenced in this episode: Selection of Controls in Case-Control Studies: I. Principles Sholom Wacholder, Joseph K. McLaughlin, Debra T. Silverman, Jack S. Mandel American Journal of Epidemiology, Volume 135, Issue 9, 1 May 1992, Pages 1019–1028, https://doi.org/10.1093/oxfordjournals.aje.a116396 Selection of Controls in Case-Control Studies: II. Types of Controls Sholom Wacholder, Debra T. Silverman, Joseph K. McLaughlin, Jack S. Mandel American Journal of Epidemiology, Volume 135, Issue 9, 1 May 1992, Pages 1029–1041, https://doi.org/10.1093/oxfordjournals.aje.a116397 Selection of controls in case-control studies. III. Design options S Wacholder 1, D T Silverman, J K McLaughlin, J S Mandel Wacholder S, Silverman DT, McLaughlin JK, Mandel JS. Selection of controls in case-control studies. III. Design options. Am J Epidemiol. 1992 May 1;135(9):1042-50. doi: 10.1093/oxfordjournals.aje.a116398

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt get some real world experience with cohort studies in a conversation with Dr. Vasan Ramachandran, PI of the Framingham Heart Study (FHS). FHS is a very well-known cohort study and the model that many of us have in mind when we think of cohort studies. We get a bit of history on FHS and Hailey and I have a chance to ask the questions we have struggled with around cohort studies including the role of representativeness. And, spoiler alert, we learn that FHS did not invent the term “risk factor” as Matt has been telling his students for years.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt get into cohort studies. We spend a lot of time confessing our limitations, both personally, and as a field, in assigning person time. We talk about the end of the large cohort study and the challenges in determining when to consider a person as exposed. We talk about issues of immortal person time and whether it is technically acceptable to include those who already have the outcome in a cohort study.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt connect with Dr. Katie Lesko for a discussion on Chapter 5 on measures of association and measures of effect. We confess our challenge with working with person time. We talk about the importance of a well specified time zero. We talk about why epidemiology is complicated by free will. We ponder what the counterfactual model looks like with time to event models. We talk about the challenges of real world data vs idealized studies. We discuss the challenges of interpreting effect measure modification. And we learn that Katie was a rower in college and is concerned that her daughter may never win an Olympic medal in gymnastics. A few papers that are mentioned in the episode: Hernán MA. Invited Commentary: Selection Bias Without Colliders. Am J Epidemiol. 2017 Jun 1;185(11):1048-1050. doi: 10.1093/aje/kwx077. PMID: 28535177; PMCID: PMC6664806. Edwards JK, Cole SR, Westreich D. All your data are always missing: incorporating bias due to measurement error into the potential outcomes framework. Int J Epidemiol. 2015 Aug;44(4):1452-9. doi: 10.1093/ije/dyu272. Epub 2015 Apr 28. PMID: 25921223; PMCID: PMC4723683. Cole SR, Hudgens MG, Brookhart MA, Westreich D. Risk. Am J Epidemiol. 2015 Feb 15;181(4):246-50. doi: 10.1093/aje/kwv001. Epub 2015 Feb 5. PMID: 25660080; PMCID: PMC4325680.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt record, then re-record due to a technical error (ooops!) a discussion on Chapter 5 on measures of association and measures of effect. We say whether we prefer risks or rates. We talk about the counterfactual, causal contrasts, valid inferences and good comparison groups. We use the phrase “living your best epi life”. And we define the difference between associations and effects. We answer whether smoking cessation programs increase the risk of being hit by a drunk driver (and if so, whether that’s causal). There is a mystery related to a mysterious death in the desert. Matt explains why he almost dropped out of intro epi. Oh and if you are wondering why this is the donut episode, Hailey sent Matt donuts after this episode after realizing (60 minutes in….) that she never pressed ‘record’ and Matt’s wife almost sent them back thinking it was a mistake since she had no idea who they were for. In the episode we mention two papers: Identifiability, exchangeability, and epidemiological confounding S Greenland, JM Robins International journal of epidemiology 15 (3), 413-419 And Confounding in health research S Greenland, H Morgenstern Annual review of public health 22 (1), 189-212

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt go back to chapter 4 of Modern Epidemiology but this time with Dr. Liz Stuart (who may not have trained as an epidemiologist but definitely thinks like an epidemiologist) who has so many insights on what seem like simple concepts. We also get into some of the differences in the way biostatisticians and epidemiologist think about these ideas. And she helps us with some of the disagreements Hailey and I had in the previous episode.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt dig into chapter 4 of Modern Epidemiology. We focused on the some of the basic building blocks of epidemiology, rates, proportions and prevalence. We found lots to discuss about defining and open and closed populations and the differences (or similarities?) between populations and cohorts. And we debate whether or not this is the “eat your vegetables” chapter. And Matt displays his ignorance of Olympic sports.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt go back to Chapters 2 and 3 of Modern Epidemiology but this time with guest Dr. Jay Kaufman of McGill University. We focused on the causal inference revolution and how our thinking on some of the issues in the chapter have changed over time as we learn more about these topics.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt try to finish off Chapter 3 of Modern Epidemiology given they couldn’t get it all into one episode as originally promised. We talked about potential outcomes, sufficient causes models and DAGs (very hard to do in audio only). We focus on the assumptions for causal inference. And we make a pitch for a Modern Epidemiology Audio Book…read by James Earl Jones.

In this episode of Season 2 of SERious Epidemiology, Hailey and Matt take on Chapters 2 and 3 of Modern Epidemiology… at least that was the plan, we really only got to chapter 2 so we’ll be back again in our next episode for Chapter 3. But in this episode we focused on some key insights around replicability and reproducibility. And camp color wars. You’ll have to listen to understand.

We are going in a new direction for Season 2 of SERious Epidemiology. This season Hailey and Matt are focusing exclusively on the new fourth edition of the textbook Modern Epidemiology. The textbook has played such an important role in the training of epidemiologists since the first edition was released and has taken on an even larger role within the field as more editions have come out. We will work through each chapter and talk about what key insights we got from them and we will talk to guests about their experiences with the text. In this first episode of the season, we are delighted to present our interview with Dr. Kenneth Rothman, author of the first edition and co-author of editions two through four. Show notes: Link to Modern Epidemiology: https://www.amazon.com/Modern-Epidemiology-Kenneth-Rothman/dp/1451193289 Link to Epidemiology: An Introduction https://www.amazon.com/Epidemiology-Introduction-Kenneth-J-Rothman/dp/0199754551/ref=sr_1_1?dchild=1&keywords=Epidemiology%3A+An+Introduction&qid=1630253351&s=books&sr=1-1