Healthy Uncertainty
The Limits of Predicting Individual-Level Longevity: Insights From the U.S. Health and Retirement Study
Luca Badolato et al.
Demography, February 2026, Pages 351-374
Abstract:
Individual-level mortality prediction is a fundamental challenge with implications for life planning, health care, social policies, and public spending. Drawing from the growing body of research on the predictability of life course events, we model and predict individual-level lifespan using 12 statistical and machine learning models and more than 150 predictors derived from the U.S. Health and Retirement Study longitudinal data. Statistical and machine learning models report comparable accuracy and relatively high discriminative performance, but they fail to account for most lifespan heterogeneity at the individual level. We observe consistent inequalities in mortality predictability and risk discrimination, with lower accuracy for men, non-Hispanic Blacks, and low-educated individuals. Additionally, people in these groups show lower accuracy in their subjective predictions of their own lifespan. Finally, top features across groups are similar, with variables related to habits, health history, and finances being relevant predictors. We conclude by highlighting the limits of predicting mortality from one of the richest longitudinal representative surveys in the United States, as well as the context-dependent inequalities across sociodemographic groups, and providing baselines and guidance for future research and public policies.
Patient and Provider Concordance: Do Patients Prefer Physicians of Their Own Race or Ethnicity?
Brigham Walker et al.
Medical Care, March 2026, Pages 161-167
Research Design: A patient-focused randomized online experiment was conducted to evaluate preferences for physicians while a physician-focused randomized field experiment was conducted to evaluate physician availability by race. The patient-focused experiment involved respondents selecting primary care physicians, while the physician-focused field experiment was conducted on a random sample of primary care physicians in Texas, which reports physician race.
Results: White respondents preferred White physicians by 10 percentage points (ppts) (95% CI: 0.048–0.157, P<0.01). Hispanic respondents favored Hispanic physicians by 27 ppts (95% CI: 0.148–0.398, P<0.01) while Black respondents favored Black physicians by 15 ppts (95% CI: −0.013 to 0.395, P=0.07). Overall, White physicians were preferred by 4.8 ppts (95% CI: 0.004–0.092, P=0.030) at the expense of Asian physicians, who were less preferred by 9.2 ppts (95% CI: −0.187 to 0.003, P=0.06). These findings are consistent with the physician-focused field experiment where Asian physicians offered appointments 3 days sooner than White providers (95% CI: −6.1 to 0.1 days, P=0.05).
Valuation of Regulatory Risk on Pharmaceutical R&D
Teresa Corzo Santamaria, Jose Portela & Eduardo Schwartz
NBER Working Paper, February 2026
Abstract:
Geopolitical tensions, supply-chain concerns and policy risk have moved to the forefront of the pharmaceutical industry. This paper develops a real options valuation model of drug R&D that captures sequential clinical investment with technical failure, stochastic costs, uncertain cash flows, and optimal abandonment. We incorporate two regulatory shocks: a reduction in effective exclusivity period and a price-control shock that reduces net cash flows. Calibrating to an incremental CNS program, we find that project value at initiation is highly right-skewed: the mean is USD 69.6m but the median is negative, so expected value is driven by rare high-upside outcomes. Regulatory risk mainly compresses this upside. Both reductions in effective exclusivity and price-based interventions substantially weaken investment incentives, even when they occur with moderate probability. Value is strongly convex in exclusivity length, with the final years carrying the highest marginal value. We introduce iso-value maps that summarize how time-based and price-based policies substitute in their impact on project valuation, to clarify the trade-offs inherent in regulatory design: losing two years of exclusivity is comparable to roughly a 30% cash-flow contraction. Using a standard revenue-to-R&D elasticity, these valuation effects imply a 10% to 25% long-run contraction in investment. The framework provides a transparent mapping from policy design to project value and investment incentives.
Tracing the Genetic Footprints of the UK National Health Service
Nicolau Martin-Bassols et al.
University of Bologna Working Paper, January 2026
Abstract:
The establishment of the UK National Health Service (NHS) in July 1948 was one of the most consequential health policy interventions of the twentieth century, providing universal and free access to medical care and substantially expanding maternal and infant health services. In this paper, we estimate the causal effect of the NHS introduction on early-life mortality and we test whether survival is selective. We adopt a regression discontinuity design under local randomization, comparing individuals born just before and just after July 1948. Leveraging newly digitized weekly death records, we document a significant decline in stillbirths and infant mortality following the introduction of the NHS, the latter driven primarily by reductions in deaths from congenital conditions and diarrhea. We then use polygenic indexes (PGIs), fixed at conception, to track changes in population composition, showing that cohorts born at or after the NHS introduction exhibit higher PGIs associated with contextually-adverse traits (e.g., depression, COPD, and preterm birth) and lower PGIs associated with contextually-valued traits (e.g., educational attainment, self-rated health, and pregnancy length), with effect sizes as large as 7.5% of a standard deviation. These results based on the UK Biobank data are robust to family-based designs and replicate in the English Longitudinal Study of Ageing and the UK Household Longitudinal Study. Effects are strongest in socioeconomically disadvantaged areas and among males. This novel evidence on the existence and magnitude of selective survival highlights how large-scale public policies can leave a persistent imprint on population composition and generate long-term survival biases.
Assessing the Effect of Deservingness Cues on Tolerance for Administrative Burdens
Simon Haeder
Policy Studies Journal, forthcoming
Abstract:
Beneficiaries of public programs must overcome several administrative challenges. Given what we know about the politics of the welfare state, it seems likely that the public's willingness to support reductions in burdens may be associated with the characteristics of potential policy targets including their life circumstances and their race/ethnicity. To learn more about how attitudes about burden reductions are affected by these factors, a survey (N = 4177) was fielded that used an experiment introducing respondents to four vignettes presenting a woman with disabilities, a single mother, an able-bodied single woman, and an individual with opioid addiction seeking enrollment in Medicaid. The experiment also used racially/ethnically identifiable names of White, Black, Hispanic, and Asian women. Respondents made clear distinctions based on individuals' life circumstances, favoring individuals with disabilities over single mothers, able-bodied single women, and individuals with opioid addiction, with the latter two consistently exhibiting the lowest levels of support. The race/ethnicity of the individual presented had very limited effects, and the effect of life circumstances consistently overshadowed those for race/ethnicity. Support for burden reductions was higher for re-enrollment than for initial enrollment. Future studies should further parse out the nuances between racial perceptions and burden tolerance.
Reducing Waste through Anti-Fraud Enforcement: Evidence from Hospital Admission Cases
David Howard & Jetson Leder-Luis
NBER Working Paper, March 2026
Abstract:
The use of federal anti-fraud laws to address unnecessary medical care is controversial. Targeted providers frequently argue that, in their judgment, the treatment in question was appropriate. We examine the effects of anti-fraud litigation against hospitals for over-admitting patients from the emergency department, using 100% Medicare claims for 2005-2019 and a design based on the staggered rollout of these lawsuits. We find that anti-fraud lawsuits reduced admission rates by 3.6 percentage points without increasing mortality rates. We estimate five-year savings to Medicare of $1.3 billion. Our results suggest that anti-fraud enforcement can be successful in reducing costly, unnecessary care.
Artificial Intelligence Length-of-Stay Forecasting and Pediatric Surgical Capacity
Jay Berry et al.
JAMA Pediatrics, March 2026, Pages 306-313
Design, Setting, and Participants: This preimplementation and postimplementation cohort study was conducted at a tertiary, freestanding, US children’s hospital among patients of any age undergoing an elective surgical procedure requiring inpatient recovery. For LOS [length-of-stay] prediction, a retrospective analysis was performed on elective surgical cases from January 1, 2018, to March 31, 2022, using Extreme Gradient Boosting (XGBoost) to predict postoperative LOS based on in-training and holdout datasets, with hyperparameter tuning using 5-fold cross-validation. For implementation and evaluation of the LOS prediction model, a preimplementation and postimplementation analysis was performed from July 1, 2022, to April 30, 2024. Data analysis was conducted from June 1 to October 31, 2024.
Results: There were 21 352 elective surgical cases (mean [SD] age, 10.2 [7.4] years; 10 804 [50.6%] female) for patients included in the retrospective analysis of postoperative LOS prediction and 12 522 elective surgical cases in the pretest and posttest analysis of the prediction model (premodel implementation, n = 5867; postmodel implementation, n = 6655). The postoperative LOS model had 85.6% accuracy with a 1-night leniency. The model’s mean absolute error was 0.6 days. After implementation of the LOS model in elective surgery scheduling and hospital bed capacity management, the median number of elective surgical procedures increased by 5 (IQR, 4.5-5) for each weekday. Variation in postoperative bedded days across days of the week decreased significantly. The magnitude of the IQR of bedded days decreased the most during midweek: 43% and 44% reductions in the IQR occurred on Wednesdays and Thursdays, respectively. The percentage of weekdays with underused capacity (<84 patients) decreased from 33% to 10% (P < .001), without a significant increase in days with excessive capacity.
A Machine Learning Model to Improve Risk Adjustment Accuracy in Medicare
Daniel Shenfeld et al.
Health Services Research, April 2026
Objective: To develop a machine learning (ML) algorithm that improves accuracy compared to the Hierarchical Condition Category (HCC) score used by the Centers for Medicare and Medicaid Services to risk-adjust payments for > 65 million Americans.
Study Design and Setting: Prognostic study using Medicare claims data to train “Franklin”, an ML algorithm predicting one-year costs, trained using identical data to HCC. Predictive accuracy was evaluated using R2 log cost, Spearman rho, and sensitivity and specificity.
Data Sources and Analytic Sample: Random sample of 2018–2019 Part A and B claims from aged, community-based enrollees in Traditional Medicare who were not dually eligible and did not have end-stage renal disease.
Principal Findings: The sample consisted of 4,176,666 Medicare beneficiaries (mean [SD] age 74.9 [7.2] years, 55.9% women; 85.9% Non-Hispanic white, 5.6% African-American, 3.4% Hispanic). Franklin was more accurate than HCC (R2 log cost 0.44 vs. 0.15; Spearman rho 0.61 vs. 0.41, p < 0.001 for both). Accuracy improved for the 47% of beneficiaries with 0 HCCs and the 27% of beneficiaries with one HCC (Spearman rho 0.59 vs. 0.08 and 0.46 vs. 0.16, respectively; p < 0.001 for both). Franklin outperformed HCC in detecting the 20% lowest-cost beneficiaries (sensitivity 0.60 vs. 0.34, specificity 0.90 vs. 0.83; p < 0.001 for both). Franklin improved accuracy over HCC for racial/ethnic minorities and rural-dwelling beneficiaries (R2 log cost Black 0.48 vs. 0.14, Hispanic 0.55 vs. 0.09, rural 0.36 v. 0.11; p < 0.001 for all), although Franklin disproportionately classified Black (15.8% vs. 10.1%) and Hispanic (22.9% vs. 12.2%) beneficiaries in the lowest predicted cost decile.
Fighting Fire with Fire: On Allowing Mergers of Large Buyers to Counter Seller Market Power
Laura Lingyu Gu et al.
University of Chicago Working Paper, February 2026
Abstract:
When dominant sellers possess substantial pricing power, buyer consolidation can serve as a countervailing force. We study this force using the 2012 merger of Express Scripts and Medco, the two largest U.S. pharmacy benefit managers (PBMs) at the time. Exploiting geographic variation in exposure to the merger, we estimate the effects of a sharp increase in buyer concentration. Prices paid to pharmacies by health insurers for branded drugs fell by about 4% in highly exposed states, with effects lasting over two years and strongest in more competitive drug-product markets. Increased buyer power also weakened the impact of seller collusion: price increases during a subsequent generic drug cartel were significantly muted in states highly exposed to the PBM merger. However, the merger also generated distortions along the supply chain. Following reductions in reimbursements, pharmacies responded by dispensing smaller package sizes, a strategy that increases revenue through more frequent refills. This response not only raised patient out-of-pocket costs, but also reduced medication adherence, which in turn resulted in subsequent increases in emergency room visits for these patients and an estimated $208 to $623 million in additional annual healthcare costs. Overall, our results highlight that buyer consolidation is double-edged: It disciplines powerful sellers, but in multi-layer supply chains it also distorts intermediary behavior in ways that can ultimately harm consumers.
The Source of Nonprofit Risk Aversion: Theory and Evidence from Hospitals
Meghan Esson, Jingshu Luo & Cameron Ellis
University of Iowa Working Paper, February 2026
Abstract:
We show that donors, not managers, drive risk-averse behavior in nonprofit organizations (NPOs). Theoretically, when donor recognition is tiered (e.g., naming rights vs. thank-you cards), donors concentrate rather than diversify giving, making shadow donation capital costs sensitive to NPO-specific risk. This induces risk-averse actions even without risk-averse managers. We test our theory using hospitals and the staggered adoption of medical liability caps and find that reduced NPO risk increases donations. Effects on substitute bond-financing indicate this increase is supply- (donor-) driven, not demand- (manager-) driven. Liability caps lead nonprofit hospitals to expand risky investments faster than for-profits, improving patient health.