Christopher F. Chabris’s research while affiliated with Geisinger Health System and other places

Genome–wide association studies (GWAS) have discovered thousands of replicable genetic associations, guiding drug target discovery and powering genetic prediction of human phenotypes and diseases. However, genetic associations can be affected by gene–environment correlations and non–random mating, which can lead to biased inferences in downstream analyses. Family–based GWAS (FGWAS) uses the natural experiment of random assignment of genotype within families to separate out the contribution of direct genetic effects (DGEs) — causal effects of alleles in an individual on an individual — from other factors contributing to genetic associations. Here, we report results from an FGWAS meta–analysis of 34 phenotypes from 17 cohorts. We found evidence that factors uncorrelated with DGEs make substantial contributions to genetic associations for 27 phenotypes, with population stratification confounding — a form of gene–environment correlation — likely the major cause. By estimating SNP heritability and genetic correlations using DGEs, we found evidence that assortative mating has led to overestimation of SNP heritability for 5 phenotypes and overestimation of the degree of shared genetic effects (pleiotropy) between 22 pairs of phenotypes. Polygenic predictors constructed from DGEs are particularly useful for studying natural selection, assortative mating, and indirect genetic effects (effects of relatives′ genes mediated through the family environment). We validate our meta–analysis results by predicting phenotypes in hold–out samples using polygenic predictors constructed from DGEs, achieving statistically significant out–of–sample prediction for 24 phenotypes with little attenuation of predictive power within–families. We provide FGWAS summary statistics for 34 phenotypes that can be used for downstream analyses. Our study provides both a template for performing FGWAS and an argument for its value for debiasing inferences and understanding the impact of environment and mating patterns.

Objectives Pragmatic randomised controlled trials (pRCTs) are essential for determining the real-world safety and effectiveness of healthcare interventions. However, both laypeople and clinicians often demonstrate experiment aversion: preferring to implement either of two interventions for everyone rather than comparing them to determine which is best. We studied whether clinician and layperson views of pRCTs for COVID-19, as well as non-COVID-19, interventions became more positive during the pandemic, which increased both the urgency and public discussion of pRCTs. Design Randomised survey experiments. Setting Geisinger, a network of hospitals and clinics in central and northeastern Pennsylvania, USA; Amazon Mechanical Turk, a research participant platform used to recruit online participants residing across the USA. Data were collected between August 2020 and February 2021. Participants 2149 clinicians (the types of people who conduct or make decisions about conducting pRCTs) and 2909 laypeople (the types of people who are included in pRCTs as patients). The clinician sample was primarily female (81%), comprised doctors (15%), physician assistants (9%), registered nurses (54%) and other medical professionals, including other nurses, genetic counsellors and medical students (23%), and the majority of clinicians (62%) had more than 10 years of experience. The layperson sample ranges in age from 18 to 88 years old (mean=38, SD=13) and the majority were white (75%) and female (56%). Outcome measures Participants read vignettes in which a hypothetical decision-maker who sought to improve health could choose to implement intervention A for all, implement intervention B for all, or experimentally compare A and B and implement the superior intervention. Participants rated and ranked the appropriateness of each decision. Experiment aversion was defined as the degree to which a participant rated the experiment below their lowest-rated intervention. Results In a survey of laypeople administered during the pandemic, we found significant aversion to experiments involving catheterisation checklists and hypertension drugs unrelated to the treatment of COVID-19 (Cohen�s d =0.25�0.46, p <0.001). Similarly, among both laypeople and clinicians, we found significant aversion to most (comparing different checklist, proning and mask protocols; Cohen�s d =0.17�0.56, p <0.001) but not all (comparing school reopening protocols; Cohen�s d =0.03, p =0.64) non-pharmaceutical COVID-19 experiments. Interestingly, we found the lowest experiment aversion to pharmaceutical COVID-19 experiments (comparing new drugs and new vaccine protocols for treating the novel coronavirus; Cohen�s d =0.04�0.12, p =0.12-0.55). Across all vignettes and samples, 28%�57% of participants expressed experiment aversion, whereas only 6%�35% expressed experiment appreciation by rating the trial higher than their highest-rated intervention. Conclusions Advancing evidence-based medicine through pRCTs will require anticipating and addressing experiment aversion among patients and healthcare professionals. Study registration http://osf.io/6p5c7/ .

The highest-achieving figures in politics, business, academia, and the media dominate public discourse and wield great influence in society. Education—perhaps especially at “elite” colleges and universities—may lie at the heart of the divide between the general public and these top achievers. In this paper, we build a new data set for the American “elite” and systematically examine the link between selective schools and outstanding achievements. In Study 1, across 30 different achievement groups totaling 26,198 people, we document patterns of attendance at a set of 34 “Elite” 34 schools, the 8 Ivy League schools, and Harvard University in particular. In Study 2, we surveyed 1810 laypeople to estimate how well they are aware of the key empirical facts from Study 1. We found that exceptional achievement is surprisingly strongly associated with “elite” education, especially obtaining a degree from Harvard, and the general public tends to underestimate the size of this effect. Attending one of just 34 institutions of higher education out of the roughly 4000 in the U.S. appears to be a critical and surprising factor separating extraordinary achievers from others in their fields.

Objective: Treatment-resistant depression (TRD) occurs in roughly one-third of all individuals with major depressive disorder (MDD). Although research has suggested a significant common variant genetic component of liability to TRD, with heritability estimated at 8% when compared with non-treatment-resistant MDD, no replicated genetic loci have been identified, and the genetic architecture of TRD remains unclear. A key barrier to this work has been the paucity of adequately powered cohorts for investigation, largely because of the challenge in prospectively investigating this phenotype. The objective of this study was to perform a well-powered genetic study of TRD. Methods: Using receipt of electroconvulsive therapy (ECT) as a surrogate for TRD, the authors applied standard machine learning methods to electronic health record data to derive predicted probabilities of receiving ECT. These probabilities were then applied as a quantitative trait in a genome-wide association study of 154,433 genotyped patients across four large biobanks. Results: Heritability estimates ranged from 2% to 4.2%, and significant genetic overlap was observed with cognition, attention deficit hyperactivity disorder, schizophrenia, alcohol and smoking traits, and body mass index. Two genome-wide significant loci were identified, both previously implicated in metabolic traits, suggesting shared biology and potential pharmacological implications. Conclusions: This work provides support for the utility of estimation of disease probability for genomic investigation and provides insights into the genetic architecture and biology of TRD.

Overconfidence is thought to be a fundamental cognitive bias, but it is typically studied in environments where people lack accurate information about their abilities. Tournament chess players receive objective, precise, and public feedback, so we conducted a preregistered survey experiment and replication to learn whether overconfidence persists in an environment that should diminish or eradicate it. Our combined sample comprised 3388 rated players aged 5–88 years, from 22 countries, with M=18.8 years of tournament experience. On average, participants asserted their ability was 89 Elo rating points higher than their observed ratings indicated—expecting to outscore an equally-rated opponent by 2:1. One year later, only 11.3% of overconfident players achieved their asserted ability rating. Low-rated players overestimated their skill the most and top-rated players were calibrated. These patterns emerged in every sociodemographic subgroup we considered. We conclude that overconfidence persists even in real-world information environments that should be inhospitable to it.

Nearly two hundred common-variant depression risk loci have been identified by genome-wide association studies (GWAS). However, the impact of rare coding variants on depression remains poorly understood. Here, we present whole-exome sequencing analyses of depression with seven different definitions based on survey, questionnaire, and electronic health records in 320,356 UK Biobank participants. We showed that the burden of rare damaging coding variants in loss-of-function intolerant genes is significantly associated with risk of depression with various definitions. We compared the rare and common genetic architecture across depression definitions by genetic correlation and showed different genetic relationships between definitions across common and rare variants. In addition, we demonstrated that the effects of rare damaging coding variant burden and polygenic risk score on depression risk are additive. The gene set burden analyses revealed overlapping rare genetic variant components with developmental disorder, autism, and schizophrenia. Our study provides insights into the contribution of rare coding variants, separately and in conjunction with common variants, on depression with various definitions and their genetic relationships with neurodevelopmental disorders.

Bipolar disorder is a leading contributor to disability, premature mortality, and suicide. Early identification of risk for bipolar disorder using generalizable predictive models trained on diverse cohorts around the United States could improve targeted assessment of high risk individuals, reduce misdiagnosis, and improve the allocation of limited mental health resources. This observational case-control study intended to develop and validate generalizable predictive models of bipolar disorder as part of the multisite, multinational PsycheMERGE Network across diverse and large biobanks with linked electronic health records (EHRs) from three academic medical centers: in the Northeast (Massachusetts General Brigham), the Mid-Atlantic (Geisinger) and the Mid-South (Vanderbilt University Medical Center). Predictive models were developed and valid with multiple algorithms at each study site: random forests, gradient boosting machines, penalized regression, including stacked ensemble learning algorithms combining them. Predictors were limited to widely available EHR-based features agnostic to a common data model including demographics, diagnostic codes, and medications. The main study outcome was bipolar disorder diagnosis as defined by the International Cohort Collection for Bipolar Disorder, 2015. In total, the study included records for 3,529,569 patients including 12,533 cases (0.3%) of bipolar disorder. After internal and external validation, algorithms demonstrated optimal performance in their respective development sites. The stacked ensemble achieved the best combination of overall discrimination (AUC = 0.82–0.87) and calibration performance with positive predictive values above 5% in the highest risk quantiles at all three study sites. In conclusion, generalizable predictive models of risk for bipolar disorder can be feasibly developed across diverse sites to enable precision medicine. Comparison of a range of machine learning methods indicated that an ensemble approach provides the best performance overall but required local retraining. These models will be disseminated via the PsycheMERGE Network website.