What We Do
CAUSALab has helped develop language and methodologies that are now standard for researchers around the world. Our work keeps pushing boundaries and innovating on the cutting-edge of causal inference methods. Our investigator team includes global experts in target trial emulation, g-methods and AI, transportability and benchmarking & others. We develop and implement these methods using real world data from electronic health records, administrative claims, biobanks, national registers, disease registries and randomized trials.
Our applied research focuses on implementing the frameworks to determine comparative effectiveness and safety of health and policy interventions. Our areas of work include infectious diseases, cardiovascular diseases, cancer, mental health & pregnancy.
VA-CAUSAL is a causal inference research initiative within the U.S. Veterans Health Administration. The goal of VA-CAUSAL is to help transform the VA into a learning health system that expedites the translation of research into practice and supports decision-making by patients, clinicians, and other stakeholders to improve health.
The Methods Core of VA-CAUSAL develops and applies causal inference methods using large-scale data resources at the VA, including electronic health records and the Million Veterans Program multi-omics biobank. Our projects include explicit emulation of target trials of sustained treatment strategies using real-world data, advanced instrumental variable estimation for Mendelian randomization, and estimation of per-protocol effects in randomized trials. The methods, computer code and materials generated during these projects are optimized for use by investigators in the VA research ecosystem and, in collaboration with the Implementation Core of VA-CAUSAL, made available to them. Our research program derives actionable insights from real-world data with timely, real-world impact.
For example, as part of our contribution to the pandemic response, we applied causal inference methods to the nationwide databases of the VA to fill evidence gaps for Covid-19 vaccines. Specifically, we led the first study to emulate a target trial of the comparative effectiveness of the BNT162b2 (Pfizer-BioNTech) and mRNA-1273 (Moderna) vaccines. We led additional studies that provided evidence for the comparative safety of these vaccines as well as the comparative effectiveness of booster doses. These studies have helped to inform vaccine decision-making at national and global levels.
-Dickerman BA, Gerlovin H, Madenci AL, Kurgansky KE, Ferolito B, Figueroa Muñiz MJ, Gagnon DR, Gaziano JM, Cho K, Casas JP, Hernán MA. Comparative effectiveness of BNT162b2 and mRNA-1273 vaccines in U.S. veterans. New England Journal of Medicine. 2021; 386(2):105-115. PMCID: PMC8693691
-Dickerman BA, Madenci AL, Gerlovin H, Kurgansky KE, Wise JK, Figueroa Muniz MJ, et al. Comparative safety of BNT162b2 and mRNA-1273 vaccines in a nationwide cohort of U.S. veterans. JAMA Internal Medicine. 2022;182(7):739-746. PMCID: PMC9194743
-Dickerman BA, Madenci AL, Gerlovin H, Kurgansky KE, Wise JK, Figueroa Muniz MJ, et al. Comparative effectiveness of third doses of mRNA-based COVID-19 vaccines in U.S. veterans. Nature Microbiology. 2023; 8(1):55-63. NIHMSID: NIHMS1864792
This work also motivated an in-depth methodological comparison of common approaches for estimating vaccine effectiveness in observational studies. Specifically, we described a methodological framework to connect the cohort design with explicit emulation of a target trial and the test-negative design, compared their performance under real-world conditions, and provided recommendations for their implementation that may serve as a guide for future studies of vaccine effectiveness.
–Li G, Gerlovin H, Figueroa Muñiz MJ, Wise JK, Madenci AL, Robins JM, Aslan M, Cho K, Gaziano JM, Lipsitch M, Casas JP, Hernán MA, Dickerman BA. Comparison of the test-negative design and cohort design with explicit target trial emulation for evaluating Covid-19 vaccine effectiveness. Epidemiology. 2023; 35(2):137-149. PMCID: PMC11022682
Lastly, this work developed the infrastructure needed to apply this causal inference framework at scale in the nationwide health care databases of the VA. This has paved the way for extensions of the framework to research across other substantive domains, including:
Cancer:
-Prostate cancer screening and risk of mortality
-Statin therapy and risk of mortality among men with prostate cancer
-Evaluating different selections and sequences of treatments for prostate cancer
Cardiovascular disease:
-Influenza vaccination and risk of cardiovascular disease
-Multifactorial Mendelian randomization for primary prevention of cardiovascular events
-Potential bias of selection into the Million Veteran Program in a Mendelian randomization study of LDL cholesterol and cardiovascular disease
-Bariatric surgery and risk of cardiovascular events
-Seasonal influenza vaccination and risk of cardiovascular events
-Duration of dual antiplatelet treatment in patients with transient ischemic attacks
Mental health:
-Statin therapy and risk of dementia
-Shingles vaccination and risk of dementia
-Lithium and risk of suicidality among patients with affective disorders
Funding:
VA-CAUSAL Methods Core is a collaboration between the Massachusetts Veterans Epidemiology Research and Information Center (MAVERIC) at the Boston VA and CAUSALab at the Harvard T.H. School of Public Health. It is funded by the Cooperative Studies Program of the U.S. Department of Veterans Affairs as part of the study CSP #2032.
Contact:
Miguel Hernán, Barbra Dickerman (Cancer), Alejandro Szmulewicz (Mental health)
CANCER-CAUSAL implements state-of-the-art causal inference methods to support cancer research.
Many important decisions about cancer control – including how to best prevent, detect, and treat cancer – must be made in the absence of evidence from randomized trials, which are often impractical or too lengthy to provide a timely answer. In these cases, we combine large health databases and sound causal inference methodology to provide the best available evidence to inform health decision-making and future trial design.
Our research program implements a systematic approach to cancer comparative effectiveness and safety research, leveraging traditional epidemiologic cohorts and novel platforms of electronic health records.
For example, in the first application of the parametric g-formula to prostate cancer research, we leveraged high-quality cohort data to provide the first survival estimates of adhering to various physical activity strategies that are sustained over time, including those recommended by current guidelines, after diagnosis with cancer.
–Dickerman BA, Giovannucci E, Pernar CH, Mucci LA, Hernán MA. Guideline- based physical activity and survival among U.S. men with nonmetastatic prostate cancer. American Journal of Epidemiology. 2019; 188(3): 579-586. PMCID: PMC6395165
In addition, we developed a causal inference framework to emulate target trials of cancer outcomes in a nationwide database of electronic health records linked to registries in the U.K. Our research demonstrated that previous randomized-observational discrepancies for statins and cancer appear to be due to the analytic approach and not any inherent problems in the observational data, and that these discrepancies disappear when observational data are analyzed using methods consistent with the causal inference framework.
-Dickerman BA, García-Albéniz X, Logan RW, Denaxas S, Hernán MA. Avoidable flaws in observational analyses: an application to statins and cancer. Nature Medicine. 2019; 25: 1601-1606. PMCID: PMC7076561
We have extended this causal inference framework to different cancer-causal questions, study designs, and data sources.
-Dickerman BA, García-Albéniz X, Logan RW, Denaxas S, Hernán MA. Emulating a target trial in case-control designs: an application to statins and colorectal cancer. International Journal of Epidemiology. 2020; 49(5): 1637-1646. PMCID: PMC7746409
-Dickerman BA, García-Albéniz X, Logan RW, Denaxas S, Hernán MA. Evaluating metformin strategies for cancer prevention: a target trial emulation using electronic health records. Epidemiology. 2023;34(5):690-9. PMCID: PMC10524586
–Guo F, McGee EE, Chiu Y, Giovannucci E, Mucci LA, Dickerman BA. Evaluating recommendation-based dietary and physical activity strategies for prostate cancer prevention: a target trial emulation in the Health Professionals Follow-up Study. American Journal of Epidemiology. 2024. Epub.
-García-Albéniz X, Hsu J, Etzioni R, Chan JM, Shi J, Dickerman BA, Hernán MA. PSA screening and prostate cancer mortality: an emulation of target trials in U.S. Medicare. JCO Clinical Cancer Informatics. 2024. In press.
Lastly, our work extends to the conduct and analysis of randomized trials, including the Nordic-European Initiative on Colorectal Cancer (NordICC) trial – a population-based trial which provided the first randomized evidence on colonoscopy screening to prevent colorectal cancer and related death. Our estimates of colonoscopy screening effectiveness and efficacy provided key evidence for priority setting on cancer screening interventions.
-Bretthauer M, Løberg M, Wieszczy P, Kalager M, Emilsson L, Garborg K, Rupinski M, Dekker E, Spaander M, Bugajski M, Holme Ø. Zauber AG, Pilonis ND, Mroz A, Kuipers EJ, Shi J, Hernán MA, Adami HO, Regula J, Hoff G, Kaminski MF. Effect of colonoscopy screening on risks of colorectal cancer and related death. New England Journal of Medicine. 2022;387(17):1547-56.
Ongoing CANCER-CAUSAL Projects include:
Prevention (Primary, Tertiary) of Cancer:
-Methodological challenges in estimating the long-term effects of lifestyle interventions for cancer survivors
-Guideline-based physical activity and quality of life among men with nonmetastatic prostate cancer: a target trial emulation in the Health Professionals Follow-up Study
Early Detection of Cancer:
-Effect of colonoscopy screening on risks of colorectal cancer and related death: instrumental variable estimation of per-protocol effects
-Prostate-specific antigen-based screening strategies and prostate cancer mortality: a target trial emulation using electronic health records
-Identifying smarter screening strategies for prostate cancer
Treatment of Cancer:
-Statin therapy for the prevention of lethal outcomes among men diagnosed with nonmetastatic prostate cancer: a target trial emulation using electronic health records
-Adjuvant bone-modifying agents and mortality among older postmenopausal women with early-stage breast cancer
-Evaluating different sequences, timing, and combinations of cancer treatments
Surveillance of Cancer:
-Comparative effectiveness of cystoscopy surveillance strategies on mortality in non-muscle invasive bladder cancer: a target trial emulation using real-world data
Funding:
CANCER-CAUSAL is funded by the U.S. National Cancer Institute (R00 CA248335; PI: Dickerman). CANCER-CAUSAL is funded by the U.S. National Cancer Institute (R00 CA248335; PI: Dickerman).
Contact:
Barbra Dickerman
The HIV-CAUSAL Collaboration is a multinational consortia of follow-up studies of persons with HIV in Europe and North America. The research team includes investigators from each of the participating studies and from the coordinating center. HIV-CAUSAL investigators love randomized trials but know that trials cannot answer all important questions for HIV clinical management in a timely manner. This is why they also use data from observational cohorts to conduct research that improves the lives of people who HIV.
For about two decades, the HIV-CAUSAL Collaboration has been at the forefront of development and implementation of methods to compare the effectiveness of clinical strategies using observational cohorts. The methodological work pioneered by the Collaboration has transformed HIV cohort analysis in the 21st century.
Besides being an incubator for new methodologies, the HIV-CAUSAL Collaboration conducts clinical research designed to inform evidence-based guidelines and the planning of randomized trials. Findings from the HIV-CAUSAL Collaboration played a prominent role in the now largely settled discussions about when to start antiretroviral therapy. Our recent research has contributed to a better understanding of the effects of antiretroviral therapy on opportunistic infections, neurological conditions, and viral resistance, as well to the study of dynamic strategies for the treatment and monitoring of individuals living with HIV.
In addition, the Collaboration facilitates understanding and training in causal modeling across leading HIV observational research groups.
HIV-CAUSAL Resources for Collaborators
Funding:
Our research is supported by the U.S. National Institute of Allergy and Infectious Diseases (R37 AI102634; PI: Hernán).
Contact:
Miguel Hernán
CAUSALab is the Methods Core of the Laboratory for Early Psychosis (LEAP) Center, one of the U.S. National Institute of Mental Health’s Advanced Laboratories for Accelerating the Reach and Impact of Treatments for Youth and Adults With Mental Illness (ALACRITY) Research Centers.
The FEP-CAUSAL Collaboration is an international consortium of prospective cohorts of individuals with first episode psychosis (FEP) that is coordinated by LEAP’s Methods Core. FEP-CAUSAL includes granular clinical and socio-demographic data, including repeated measures of symptom severity and psychosocial functioning, on more than 2500 individuals recovering from a first episode of psychosis from across the globe. The goal of FEP-CAUSAL is to study interventions to guide treatment choices of clinicians and patients with first episode of psychosis. The first scientific report using FEP-CAUSAL data estimated the comparative effectiveness of different antipsychotic agents for initial choice as treatment for FEP, successfully benchmarking estimates with the largest RCT of antipsychotic initiation in FEP to date and extending the results to new medications (Szmulewicz et al, Am J Epidemiol 2023). Current ongoing projects aim to examine gender differences in clinical trajectories, treatments, and outcomes of FEP patients; the comparative effectiveness of different antipsychotic treatments in terms of psychosocial functioning; the potential role of olanzapine use for treatment-resistant FEP patients; and medication discontinuation strategies for patients with sustained clinical recovery.
Funding:
Our research is supported by the U.S. National Institute of Mental Health (grant P50 MH115846 (PIs: Öngür, Hsu, Hernán) and by the Brain & Behavior Research Foundation (Grant 2022-31313; PI: Szmulewicz).
Contact:
Alejandro Szmulewicz
Hep-CAUSAL is a consortium of cohorts of people with HIV and HCV co-infection in Europe and North America. The research team includes investigators from each of the participating studies and from the coordinating center.
HIV and HCV co-infection places patients at risk of morbidity and mortality from hepatic disease (cirrhosis, hepatocellular carcinoma) and extra-hepatic morbidities (cardiovascular disease, kidney disease, diabetes). Treatment with direct-acting antiviral agents (DAA) cures HCV infection in about 95% of patients. However, even after HCV cure is achieved, hepatic and extra-hepatic disease remains risks, and some individuals become reinfected with HCV.
By combining observational data and cutting-edge causal inference methods, HepCAUSAL estimates the long-term risk of HCV reinfection, hepatic disease, and extra-hepatic disease under current guidelines that recommend DAA treatment for all persons with HIV and HCV co-infection. As millions of people are expected to receive DAA treatment in the US and globally, our findings inform clinical guidelines and optimal screening strategies and how to maximize the benefits of DAA treatment in the long-term.
Hep-CAUSAL Resources for Collaborators
Funding:
Our research is supported by the U.S. National Institute of Allergy and Infectious Diseases (R37 AI102634; PI: Hernán) and the Providence/Boston Center for AIDS Research (P30AI042853).
Contact:
Sara Lodi
CAUSALab investigators have pioneered the application of the target trial framework to the generation of evidence on the safety and effectiveness of interventions during pregnancy. CAUSALab works in partnership with H4P, The Harvard Program on Perinatal and Pediatric Pharmacoepidmiology, which includes investigators from Brigham and Women’s Hospital, Harvard Medical School, and the Harvard T.H. Chan School of Public Health, learn more about the team.
The H4P team works to fill the gap of information on the effects of drugs and vaccines on the mother and the developing fetus. Through this work, they are determined to improve clinical decisions and improve health outcomes for this historically underserved population. Learn more about their research.
Funding:
Our research is supported by grant R01 HD097778 (PI: Hernández-Díaz) from the National Institute of Child Health and Human Development, U.S. National Institutes of Health.
Contact:
Sonia Hernández-Díaz
Come back soon for more information about research on benchmarking and transportability.