In 1986, Professor James Robins described a generalized theory of causal inference from complex longitudinal data with time-varying treatments. This seminal paper marked the beginning of an era in causal inference research from randomized and observational studies. Over the next decades, and under Robins’s scientific guidance, our group in the Department of Epidemiology at the Harvard T.H. Chan School of Public Health made groundbreaking contributions to methodology for causal inference. We applied our methods to address important questions that span a wide range of health areas. In 2021, the CAUSALab was founded under the direction of Miguel Hernán to articulate a growing research portfolio, create synergy with our strategic partners, and provide training on causal inference to the next generation of investigators.
Kolokotrones Symposium
The Kolokotrones Symposium on Data Science is a monthly gathering focused on discussing methodologic issues that arise in data science, including epidemiology and biostatistics, in a more relaxed setting, These monthly meetings are organized as methods clinics during which students, postdocs, and other junior investigators present questions about methodological challenges that the encounter in their research. Possible solutions are discussed by the group, which includes faculty members from the Departments of Biostatistics and Epidemiology. The discussion topics can encompass any aspect of data science, including database management, design of observational analyses, machine learning algorithms, and causal inference techniques.
Do you have ideas for future Kolokotrones symposium topics? If so, we would love to hear from you! Please send any thoughts to Mats J. Stensrud (mstensrud@hsph.harvard.edu).
Upcoming Symposium
Details regarding the next Kolokotrones Symposium will be released at a later date.
Past Symposia
Spring 2020
Weighting for Causal Inference: Estimating the impact of a natural disaster on health outcomes?
February 7, 2020 | 3:45-5:15pm | TMEC 227 Amphitheater
Hosts: José R. Zubizarreta
Discussant: Jamie Robins
Moderator: Miguel Hernan
Sponsored by the Program on Causal Inference
Fall 2019
Bugs and drugs: How (and how much) does changing antibiotic use, change antibiotic resistance?
November 1, 2019 | 3:45-5:15pm | 503, Ballard Room
Co-Hosts: Marc Lipsitch and Yonatan Grad
Moderator: Mats J. Stensrud
Sponsored by the Program on Causal Inference
Mobile Health: Mobile Health: Which type of in-the-moment support is best provided
to someone who is struggling with suicidal thoughts?
October 4, 2019 | 3:45-5:15pm | 503, Ballard Room
Co-Hosts: Susan Murphy and Matt Nock
Moderator: Miguel Hernan
Sponsored by the Program on Causal Inference
Spring 2019
Maching Learning for Health Sciences: How can we better leverage machine learning to change the future of healthcare?
Friday May 3, 2019 | 3:45-5:15pm | Waterhouse Faculty Room, 106 Gordon Hall
Co-Hosts: Andrew Beam, Sherri Rose, and David Sontag
Introduction by Michael Oberst
Dr. Beam is a Senior Fellow at Flagship Pioneering, the head of machine learning at a Flagship-backed startup, and a visiting lecturer in the Department of Epidemiology at the HSPH. His research solves problems in medicine through the use of machine learning, and he has a special interest in developing models for preterm infants in the neonatal intensive care unit.
Dr. Rose is an associate professor in the Department of Health Care Policy at the Harvard Medical School and the co-director of the Health Policy Data Science Lab, whose methodological research focuses on statistical machine learning with applications in health policy.
Dr. Sontag is an associate professor of electrical engineering and computer science at MIT, who focuses on advancing machine learning and artificial intelligence to address issues in health care.
Moderator: Miguel Hernan
Sponsored by the Program on Causal Inference
Click here to view the May 2019 flyer.
What Problems Can Electronic Health Records Solve? The strengths and pitfalls of EHR data
Friday April 5, 2019 | 3:45-5:15pm | Ballard Room, Countway Library
Co-Hosts: James Floyd, Jessica Chubak, and Goodarz Danaei
Dr. Floyd is an Assistant Professor in the Departments of Medicine and Epidemiology at the University of Washington, an Affiliate Investigator at Kaiser Permanente Washington Health Research Institute, and practices medicine as a general internist at Harborview Medical Center in Seattle, WA.
Dr. Chubak is a Senior Investigator at the Kaiser Permanente Washington Health Research Institute, and an Affiliate Associate Professor in the Department of Epidemiology at the University of Washington.
Dr. Danaei is an Associate Professor of Global Health in the Departments of Epidemiology and Global Health and Population at the Harvard T.H. Chan School of Public Health.
Moderator: Miguel Hernan
Sponsored by the Program on Causal Inference
Click here to view the April 2019 flyer.
Do the results from a trial apply to my target population? Transporting causal inferences from randomized trials to new populations
Friday March 8, 2019 | 3:45-5:15pm | 106 Gordon Hall
Discussion jump-started by: Elvira D’Andrea, Irina Degtiar, and Sarah Robertson
Co-Hosts: Issa Dahabreh and Mehdi Najafzadeh
Moderator: Miguel Hernan
Co-sponsored by the Center for Evidence Synthesis in Health at Brown University, the Division of Pharmacoepidemiology & Pharmacoeconomics at Brigham and Women’s Hospital, and the HSPH Program on Causal Inference
Click here to view the March 2019 flyer.
Using Real-World Data to Emulate Randomized Trials – Target Trial Design and Implementation
Friday February 1, 2019 | 3:45-5:15pm | 106 Gordon Hall
Heard about target trials and curious to learn more? Join us for a collaborative group exercise in designing target trials! Using examples from pharmacoepidemiology and clinical epidemiology, this workshop will provide a practical tutorial on conceiving and designing target trials, and avoiding common pitfalls encountered when using that target trial to guide study design and analyses in observational data. No prior experience with target trials, pharmacoepidemiology, or clinical epidemiology required.
Facilitators: Barbra Dickerman, Lucia Petito, & Kerollos Wanis
Hosts: Sara Lodi & Miguel Hernan
Co-sponsored by the HSPH Program on Causal Inference and Boston University School of Public Health Department of Biostatistics
Click here to view the February 2019 flyer.
Fall 2018
Workshop: Crowdsourcing Causal Graphs – A hackathon for causal DAGs
Friday December 7, 2018 | 3:45-5:15pm | 261 Tosteson Medical Education Center – 2nd Floor Atrium
Heard about causal graphs or DAGs and curious to know more? Join us for a collaborative, hands-on exercise in building causal graphs! Using an example from psychiatric epidemiology, this workshop will provide a practical tutorial on summarizing expert knowledge into a causal graph and using that graph to think about study design and analyses. No prior experience with causal graphs or psychiatric epidemiology required.
Facilitators: Eleanor Murray, Jaimie Gradus, & Matthew Fox
Presenters: Louisa Hills Smith & Laura Ann Sampson
Host: Sonia Hernandez-Diaz
Co-sponsored by the HSPH Program on Causal Inference and Boston University School of Public Health Department of Epidemiology
Click here to view the December 2018 flyer.
Workshop: Causal Survival Analysis in Follow-Up Studies
Friday November 2, 2018 | 3:45-5:30pm | Ballard Room, Countway Library
Do you work with survival data? Are you wary of the hazards of hazard ratios? Learn how to estimate survival outcomes adjusted for baseline and time-varying confounders! This practical, hands-on workshop will cover estimating survival curves, cumulative incidence differences & ratios, and inverse probability-weighted hazard ratios for intention-to-treat effects and per-protocol treatment effects for randomized trials and observational data. Bring a laptop with R or SAS, we’ll provide the code and real data!
Facilitated by Lucia Petito & Eleanor Murray
Hosted by Miguel Hernán
Click here to view the November 2018 flyer.
Should Women Use Acetaminophen during Pregnancy? A discussion of analytic choices and improbable study findings.
Friday October 5, 2018 | 3:45-5:30pm | Ballard Room, Countway Library
Currently, acetaminophen is the only over-the-counter pharmaceutical drug approved by the Food and Drug Administration (FDA) for women to use during pregnancy for pain relief or fever reduction. However, recent studies done in cohorts around the world have found an increased risk of neurocognitive outcomes (e.g. ADHD diagnosis, slower language development, lower IQ) in children of women who used acetaminophen during pregnancy. In 2014, the FDA reviewed this research and deemed the level of evidence not high enough to change the current recommendation. In this symposium, we will discuss possible sources of bias in these analyses of cohort data that may have influenced study findings (unmeasured confounding, measurement error, selection bias), what the regulatory implications of these findings are, and how to proceed when study findings polarize the scientific community.
Presentations by: Samantha Parker, Mollie Wood, & Yanmin Zhu
Panel members: Miguel Hernán (moderator), Sonia Hernandez-Diaz, Samantha Parker, Martha Werler, Mollie Wood, & Yanmin Zhu
Click here to view the October 2018 flyer.
Spring 2018
The Data Science of Implementation Science
Friday, May 4, 2018 | 3:45-5:30pm | Ballard Room, Countway Library
Implementation science aims to study the process by which original research becomes widely disseminated. It has been described as “a systematic, scientific approach to ask and answer questions about how to get ‘what works’ to people who need it with greater speed, fidelity, efficiency, quality, and relevant coverage.” However, there are many analytic challenges to consider when assessing the effectiveness of the implementation of a public health intervention, which may, at many stages, not be what the original research recommended. In this symposium, we will 1) describe the basic principles of implementation science, 2) provide a concrete example of the public health significance of an implementation science study, and 3) demonstrate some of the methodological work underway to address such challenges.
Presentations by: Dale Barnhart, John Naslund, & Daniel Nevo
Panel members: Dale Barnhart, Miguel Hernán, John Naslund, Daniel Nevo, Lucia Petito (moderator), & Donna Spiegelman
Click here to view the May 2018 flyer
Did your mother’s prenatal diet affect you? Using lifecourse data to answer intergenerational questions
Friday, April 6, 2018 | 3:45-5:30pm | Ballard Room, Countway Library
Co-hosted by the Division of Chronic Disease Research Across the Lifecourse (CoRAL) and the Program on Causal Inference
Many chronic childhood diseases are thought to originate in the early stages of human development. Prenatal diet is thought to be a modifiable risk factor that directly impacts child growth in utero. In this symposium, we plan to discuss 1) the evidence that supports the effect of prenatal diet on offspring childhood health outcomes, 2) how researchers use complex longitudinal surveys to conduct intergenerational research, and 3) how to incorporate causal inference ideas into future analyses.
Presentations by: Izzuddin Aris, Yu-Han Chiu, & Karen Switkowski
Panel members: Izzuddin Aris, Yu-Han Chiu, Miguel Hernán, Emily Oken, Karen Switkowski, & Jessica Young (moderator)
Click here to view the April 2018 flyer
Evidence-Based Policy: How can we use health data to build better microsimulation models?
Friday, March 2, 2018 | 3:45-5:30pm | Ballard Room, Countway Library
Co-hosted by the Center for Health Decision Science and the Program on Causal Inference
Many health policy questions cannot be answered by conducting randomized trials or analyzing observational data. To address these questions, researchers have turned to microsimulation models, which incorporate data from varying sources. In this symposium, we aim to better understand: 1) the types of questions microsimulation models are used to answer, 2) the evidence used to create an individual-level microsimulation model, and 3) the connections between microsimulation models and the parametric g-formula, a causal inference technique. We will discuss recommendations for practical implementation of both methodologies.
Presentations by: Nicole Campos, Vidit Munshi, & Eleanor Murray
Panel members: Nicole Campos, Miguel Hernán, Jane Kim, Vidit Munshi, Eleanor Murray, & Steve Resch (moderator)
Click here to view the for March 2018 flyer
Do statins increase the survival of cancer patients?
Friday, February 2, 2018 | 3:45-5:30pm | Ballard Room, Countway Library
Co-hosted by the Program on Causal Inference, the Cancer Epidemiology Program, and the Harvard Chan Student Club on CANcer: Learn, Interact, Translate (CAN-LIT)
Over 25 million Americans take statins. While the cardiovascular benefits of statins are not under discussion, there is a debate as to whether statins increase the overall survival of cancer patients. This protective effect is biologically plausible and, in fact, reports from several observational studies have found an inverse association between statins and mortality among cancers patients. However, some of these studies did not explicitly emulate a randomized trial, which might have introduced selection bias and immortal time bias in the estimates. So do statins increase survival in cancer patients or is the inverse association due to bias? In this symposium we will address this question by discussing the available evidence on statins and mortality in cancer patients and will chart a research path forward.
Presentations by: Louise Emilsson, Hari Iyer, & Claire Pernar
Panel discussion: Barbra Dickerman (moderator), Louise Emilsson, Miguel Hernán, Hari Iyer, Kevin Kensler (moderator), Lorelei Mucci, Claire Pernar, & Joy Shi (moderator)
Kolokotrones Circle
The Kolokotrones Circle Meetings are weekly gatherings focused on discussing methodologic issues that arise in data science and causal inference. These weekly meetings are organized as highly informal presentations, during which students, postdocs, and other junior investigators present issues encountered in the course of their ongoing research. Possible solutions are discussed by the group, which includes faculty members from the Departments of Biostatistics and Epidemiology. The discussion topics encompass all aspects of data science, including database management, design of observational analyses, machine learning algorithms, and causal inference techniques. All meetings are on Fridays from 3:45-5:15pm in Kresge 201 unless otherwise stated.
The Kolokotrones Circles are canceled until further notice due to the COVID19 pandemic.