Medical Conditions
The Effect of On-the-Job Experience on Base-Rate Neglect: Evidence from Medical Professionals
Felipe Araujo, Christopher Chabris & Michelle Meyer
Management Science, forthcoming
Abstract:
We study the effect of on-the-job experience on base-rate neglect, which is a common bias in assessing conditional probabilities. We do so by carrying out experiments with medical professionals, who are routinely exposed to conditional-probability problems in the form of diagnostic tests, and nonmedical professionals, who are not. As such, medical workers with more years of experience will have had more exposure to base-rate type problems than nonmedical workers with similar years of experience. We estimate the effect of on-the-job experience by comparing the answers of more or less experienced professionals in both the medical and nonmedical domains. Although the incidence of the bias is high for both groups and all levels of experience, we find that more experienced medical workers (a) have lower rates of perfect base-rate neglect (i.e., completely ignoring the base rates), (b) provide more accurate posterior estimates, and (c) adjust their estimates more in response to changes in the base rates. We observe no such difference for nonmedical workers. We conduct a number of robustness checks and consider possible mechanisms, such as education, job or survey attrition, selectivity into medical professions, and experience with false positives. Our results suggests that on-the-job experience mitigates, but does not eliminate, base-rate neglect.
Race to the Bottom: Competition and Quality in Science
Ryan Hill & Carolyn Stein
Quarterly Journal of Economics, forthcoming
Abstract:
This paper investigates how competition to publish first and thereby establish priority impacts the quality of scientific research. We begin by developing a model where scientists decide whether and how long to work on a given project. When deciding how long they should let their projects mature, scientists trade off the marginal benefit of higher quality research against the marginal risk of being preempted. Projects with the highest scientific potential are the most competitive because they induce the most entry. Therefore, the model predicts these projects are also the most rushed and lowest quality. We test the predictions of this model in the field of structural biology using data from the Protein Data Bank (PDB), a repository for structures of large macromolecules. An important feature of the PDB is that it assigns objective measures of scientific quality to each structure. As suggested by the model, we find that structures with higher ex-ante potential generate more competition, are completed faster, and are lower quality. Consistent with the model, and with a causal interpretation of our empirical results, these relationships are mitigated when we focus on structures deposited by scientists who -- by nature of their employment position -- are less focused on publication and priority. We estimate that the costs associated with improving these low-quality structures are between 1.5 and 8.8 billion dollars since the PDB’s founding in 1971.
Algorithmic Recommendation Tools and Experiential Learning in Clinical Care
Shirish Sundaresan & Isin Guler
Organization Science, forthcoming
Abstract:
This study examines the relationship between the adoption of algorithmic recommendation tools and experiential learning. We argue that the adoption of an algorithmic recommendation tool will harm experiential learning in organizations by limiting knowledge retention and retrieval. We further argue that the adverse relationship between algorithmic tool adoption and experiential learning will be stronger in organizations operating in low-task-difficulty environments than those in high-task-difficulty ones because organizational members in such organizations are likely to rely more on algorithmic recommendations, experiencing higher skill erosion. In addition, the relationship will be stronger in organizations facing low task variety than in those with high task variety, as these organizations are likely to have more rigid routines and in turn experience higher routine disruption after adopting an algorithmic tool. We utilize data on the adoption of an algorithmic tool called a clinical decision support system (CDSS) in a sample of emergency departments in California and utilize a fixed-effects panel regression with control function to test our arguments. We find that the relationship between cumulative experience and mortality becomes significantly weaker after CDSS adoption, suggesting flatter learning curves. We also find evidence that the effect is moderated by task difficulty and task variety.
Clonal Candida auris and ESKAPE pathogens on the skin of residents of nursing homes
Diana Proctor et al.
Nature, forthcoming
Abstract:
Antimicrobial resistance is a public health threat associated with increased morbidity, mortality and financial burden in nursing homes and other healthcare settings. Residents of nursing homes are at increased risk of pathogen colonization and infection owing to antimicrobial-resistant bacteria and fungi. Nursing homes act as reservoirs, amplifiers and disseminators of antimicrobial resistance in healthcare networks and across geographical regions. Here we investigate the genomic epidemiology of the emerging, multidrug-resistant human fungal pathogen Candida auris in a ventilator-capable nursing home. Coupling strain-resolved metagenomics with isolate sequencing, we report skin colonization and clonal spread of C. auris on the skin of nursing home residents and throughout a metropolitan region. We also report that most Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Entobacter species (ESKAPE) pathogens and other high-priority pathogens (including Escherichia coli, Providencia stuartii, Proteus mirabilis and Morganella morganii) are shared in a nursing home. Integrating microbiome and clinical microbiology data, we detect carbapenemase genes at multiple skin sites on residents identified as carriers of these genes. We analyse publicly available shotgun metagenomic samples (stool and skin) collected from residents with varying medical conditions living in seven other nursing homes and provide additional evidence of previously unappreciated bacterial strain sharing. Taken together, our data suggest that skin is a reservoir for colonization by C. auris and ESKAPE pathogens and their associated antimicrobial-resistance genes.
The Value of Specific Knowledge: Evidence from Disruptions to the Patient–Physician Relationship
Stephen Schwab
Management Science, forthcoming
Abstract:
When a member of a work team leaves, some knowledge is lost to the organization. Exploiting quasi-random turnover among military physicians because of deployments, I estimate the effects of turnover on patients and other providers in the same care team. I find that a discontinuity in primary care leads to a 3%-5% increase in costs driven primarily by an increase in the use and intensity of specialty care with no observable benefit to the patient as measured by potential reductions in hospitalization rates and emergency department usage. This indicates that the full cost of turnover includes a reduction in access to knowledge among remaining members of the team.
Limited diffusion of scientific knowledge forecasts collapse
Donghyun Kang et al.
Nature Human Behaviour, February 2025, Pages 268-276
Abstract:
Market bubbles emerge when asset prices are driven unsustainably higher than asset values, and shifts in belief burst them. We demonstrate an analogous phenomenon in the case of biomedical knowledge, when promising research receives inflated attention. We introduce a diffusion index that quantifies whether research areas have been amplified within social and scientific bubbles, or have diffused and become evaluated more broadly. We illustrate the utility of our diffusion approach in tracking the trajectories of cardiac stem cell research (a bubble that collapsed) and cancer immunotherapy (which showed sustained growth). We then trace the diffusion of 28,504 subfields in biomedicine comprising nearly 1.9 M papers and more than 80 M citations to demonstrate that limited diffusion of biomedical knowledge anticipates abrupt decreases in popularity. Our analysis emphasizes that restricted diffusion, implying a socio-epistemic bubble, leads to dramatic collapses in relevance and attention accorded to scientific knowledge.
Referral Triads
Mathijs de Vaan & Toby Stuart
Administrative Science Quarterly, March 2025, Pages 157-193
Abstract:
Third parties who refer clients to expert service providers help clients navigate market uncertainty by curating well-tailored matches between clients and experts and by facilitating post-match trust. We argue that these two functions often entail trade-offs because they require referrers to activate network relationships with different experts. While strong ties between referrers and experts promote trust between clients and experts, the presence of such ties reduces the likelihood that intermediaries refer clients to socially distal experts who may be better suited to serve clients’ needs. We examine this central and unexplored tension by using full population medical claims data for the state of Massachusetts. We find that when primary care physicians (PCPs) refer patients to specialists with whom the PCPs have strong ties, patients demonstrate more confidence in the specialists’ recommendations. However, a strong tie between the PCP and specialist also reduces the expertise match between a patient’s health condition and a specialist’s clinical experience. These findings suggest that the two central means by which referrers add value may be at odds with one another because they are maximized by the activation of different network ties.