System of Care

Kevin Lewis

July 12, 2021

Swarm Learning for decentralized and confidential clinical machine learning
Stefanie Warnat-Herresthal et al.
Nature, 10 June 2021, Pages 265–270


Fast and reliable detection of patients with severe and heterogeneous illnesses is a major goal of precision medicine. Patients with leukaemia can be identified using machine learning on the basis of their blood transcriptomes. However, there is an increasing divide between what is technically possible and what is allowed, because of privacy legislation. Here, to facilitate the integration of any medical data from any data owner worldwide without violating privacy laws, we introduce Swarm Learning -- a decentralized machine-learning approach that unites edge computing, blockchain-based peer-to-peer networking and coordination while maintaining confidentiality without the need for a central coordinator, thereby going beyond federated learning. To illustrate the feasibility of using Swarm Learning to develop disease classifiers using distributed data, we chose four use cases of heterogeneous diseases (COVID-19, tuberculosis, leukaemia and lung pathologies). With more than 16,400 blood transcriptomes derived from 127 clinical studies with non-uniform distributions of cases and controls and substantial study biases, as well as more than 95,000 chest X-ray images, we show that Swarm Learning classifiers outperform those developed at individual sites. In addition, Swarm Learning completely fulfils local confidentiality regulations by design. We believe that this approach will notably accelerate the introduction of precision medicine.

Association of Medicaid Expansion with Suicide Deaths among Nonelderly U.S. Adults
Anna Austin, Rebecca Naumann & Nicole Short
American Journal of Epidemiology, forthcoming


In 2014, the Affordable Care Act gave states the option to expand Medicaid coverage to nonelderly adults (18-64 years) with incomes up to 138% of the federal poverty level. The association of Medicaid expansion with suicide, a leading cause of death in the U.S., has not been examined. We used 2005-2017 National Violent Death Reporting System data for eight Medicaid expansion and seven non-expansion states. We examined the association of Medicaid expansion with rates of suicide death among nonelderly adults per 100,000 population using a difference-in-differences approach. Adjusting for state-level confounders, Medicaid expansion states had 1.2 (95% CI -2.5, 0.1) fewer suicide deaths per 100,000 population per year in the post-expansion period than would have been expected if they had followed the same trend in suicide rates as non-expansion states. Medicaid expansion was associated with reductions in suicide rates among women, men, those 30-44 years, white, non-Hispanic individuals, and those without a college degree. Medicaid expansion was not associated with a change in suicide rates among those 18-29 or 45-64 years, and non-white or Hispanic individuals. Overall, Medicaid expansion was associated with reductions in rates of suicide death among nonelderly adults. Further research on inequities in expansion benefits is needed.

Moving to Better Healthcare? Estimating the Causal Impact of Medicaid Expansion on Homelessness
Ashutosh Kumar
Washington State University Working Paper, April 2021


Recent homelessness figures in the United States show a puzzling trend of aggregate decline but lopsided changes across states. This study examines the puzzle through healthcare access and provides the first causal evidence of the Medicaid expansion’s impact on homeless adults’ location. Using the state and county-level data on the homeless population from 2010-2017, the estimates from a difference-in-differences model show a significant 10 % post-expansion increase in homeless individuals per capita in states that adopted Medicaid expansion. Furthermore, utilizing the difference in homeless individuals’ coverage status vis-a-vis homeless people in families, estimates from a triple difference (DDD) model also confirm the post-expansion increase in homelessness in expansion states. This study contributes further by uniquely utilizing county-level data to provide subsample analysis on metropolitan counties and counties located at state borders. Results from the state, county, and border-county-discontinuity design reveal the evidence of homeless individuals’ migration from non-expansion to expansion states. Two mechanisms explain the migration process: post-expansion coverage eligibility of previously uninsured homeless individuals and the increased ability of homeless service providers in expansion states in offering healthcare and housing-related services. This paper concludes by measuring the state spending on Medicaid to demonstrate the implications of these findings on state welfare policymaking and fiscal expenditure.

Connecting Medicaid and child support: Evidence from the TennCare disenrollment
Lindsey Rose Bullinger & Sebastian Tello-Trillo
Review of Economics of the Household, September 2021, Pages 785–812


In 2005, Tennessee disenrolled about 170,000 adults from its Medicaid program. This health insurance reform has been shown to decrease access to care and worsen financial well-being. Those affected by the disenrollment greatly overlaps with the population responsible for paying child support. We examine whether the contraction of public health insurance eligibility in Tennessee affected child support receipt. We find that after TennCare disenrollment, parents with custody of children in Tennessee were 7.2 percentage points (16 percent) less likely to receive child support income. Child support income also decreased by $204 (20 percent). We document heterogenous effects, with the largest effect among custodial parents who are non-white, under age 40, and above the poverty line.

Heterogeneity in the Impact of Privatizing Social Health Insurance: Evidence from California's Medicaid Program
Mark Duggan, Craig Garthwaite & Adelina Yanyue Wang
NBER Working Paper, June 2021


State governments face the classic “make or buy” decision for the provision of Medicaid services. Over the past two decades, the majority of states have outsourced the provision of social health insurance through Medicaid Managed Care (MMC) programs. These programs have been extensively studied in the literature – with little evidence of large positive or negative effects. However, most states initially allowed older and sicker enrollees to remain enrolled in the government run fee for service (FFS) programs. It is possible that these more fragile enrollees could have a different experience in managed care. In this paper we study California’s mandatory enrollment of the senior and persons with disabilities (SPD) population in MMC. We find this mandatory enrollment caused an increased use of the emergency department and transfers between hospitals. This was not simply a hassle cost for enrollees – we also estimate an increase in mortality for the affected population. These effects were strongest for the sickest enrollees – the types of enrollees that might be expected to have a different experience with managed care. Our results suggest the adverse impact of MMC varies by the enrollee health, which should inform the optimal outsourcing decision for governments.

Reverse Cross Subsidization in Healthcare Capitation Programs: Evidence from Medicare Advantage
Zhaowei She et al.
Georgia Tech Working Paper, May 2021


Capitation payment models have been increasingly adopted by the payers in the U.S. healthcare market during the past decade. However, healthcare services provided in Medicare Advantage (MA), the largest capitation program in the U.S., have been suggested to be more appealing to healthier patients and less appealing to sicker patients. The mismatch between a patient's health status and the benefits she gets from MA suggests that there may be a misallocation problem in MA. Despite extensive research on Medicare capitation program, little is known about how MA health plans actually allocate these capitation payments to different patients due to limited access to MA health plans' claims data. This paper utilizes a large commercial insurance database containing claims from more than 2 million MA enrollees to study the allocation problem of MA capitation payments. We empirically demonstrate that MA inadvertently incentivizes MA health plans to reallocate parts of the capitation payments from the sick to cross subsidize the healthy. By exploiting an exogenous policy shock on MA capitation payments through a Difference-in-Difference (DID) design, we identify, the first time in the literature, this reverse cross subsidization practice. Furthermore, we show that the reverse cross subsidization practice is associated with the risk selection problem in MA, where low-risk patients are more likely to enroll in MA compared to the high-risk patients.

Are drug prices subject to creative destruction? Evidence from the US, 1997–2017
Frank Lichtenberg
Health Economics, August 2021, Pages 1910-1932


There are several types of pharmaceutical competition. In addition to competition among producers of the same chemical substance (“within-substance competition”), there may be competition among producers of different chemical substances in the same chemical subgroup (“between-substance competition”). There have been numerous econometric studies of the effect of within-substance competition on drug prices, but empirical evidence about the effect of between-substance competition is far more limited. The primary objective of this study is to assess the impact of the entry of new drugs in a drug's therapeutic class on branded drug prices, generic drug prices, and the generic market share, using publicly-available US data for the period 1997–2017. Two methods are used to estimate the effects of between-substance and within-substance competition on those variables. The first method is standard 2-way fixed effects estimation based on aggregate data. The second method, based on micro data, is estimation using the DID_MULTIPLEGT procedure developed by de Chaisemartin et al. (2021), which does not rely on, and allows us to test for, “parallel trends.” Between-substance competition does not appear to have any effect on brand-name drug prices, although our inability to fully account for rebates may bias the estimates towards zero. (There is also little evidence for an effect of within-substance competition on brand-name drug prices.) However, between-substance competition has a significant negative effect on generic drug prices. We estimate that the 1985–2005 increase in the number of substances ever registered in a drug's ATC4 chemical subgroup reduced the 2017 price of generic drugs by 42%. (The ratio of the generic-price reduction attributable to rising between-substance competition to the generic-price reduction attributable to rising within-substance competition also happens to be 42%.) A striking finding is that the entry of imitators has no effect on the prices of brand-name drugs, but the entry of innovators has a significant negative effect on the prices of generic drugs in the same ATC4 chemical subgroup. In addition, between-substance competition has a significant positive effect on the generic market share: the 1985–2005 increase in the number of substances ever registered in a drug's ATC4 chemical subgroup increased the 2017 generic market share by 15.0 percentage points. Due to its effects on generic drug prices and the generic market share, the 1985–2005 increase in between-substance competition reduced the average 2017 price of drugs that were already sold in 1997 by 35%. We estimate that 36% of 2017 expenditure on drugs that were first registered during 1986–2005 was offset by reduced 2017 expenditure on drugs that were sold in both 1997 and 2017.

Regulatory Approval and Expanded Market Size
Benjamin Berger, Amitabh Chandra & Craig Garthwaite
NBER Working Paper, June 2021


Regulatory review of new medicines is often viewed as a hindrance to innovation by increasing the hurdle to bring products to market. However, a more complete accounting of regulation must also account for its potential market expanding effects through quality certification. We combine data on FDA approvals for follow-on indications and patient-level data on utilization, and examine whether FDA approval of a follow-on indication increases the use of a drug for that indication. We find 5 facts for the market-expanding role of regulation: (1) follow-on approvals increase the share of patients taking a drug with that indication by 4.1 percentage points, or 40% increase over baseline use, at the time of approval; (2) there is little market learning prior to or following the approval of the follow-on indication, suggesting that such approvals fully certify the new use; (3) the effect of these approvals is larger for uses in a different disease area than previous indications, an increase equivalent to over 4 years of market-learning; (4) it is FDA approval, not the initiation of clinical trials that generate the expansion in market size; (5) the market expansion is consistent with physicians prescribing the medicines more because of higher perceived benefits, not reduced administrative costs.

High-frequency hospital users: The tail that wags the readmissions dog
Hassan Fouayzi & Arlene Ash
Health Services Research, forthcoming

Data sources: 2015 and 2016 Massachusetts Medicaid data.

Study design: We compare demographics, morbidity burden, and social risk factors for high- and low-frequency hospital users, and membership in 17 accountable care organizations. We evaluate how excluding hospitalizations of high-frequency users from a 30-day readmission measure (with or without risk adjustment) changes its rate and variability and affects performance rankings of accountable care organizations. The outcome is readmission within 30 days; each live discharge from a hospital contributes one observation.

Principal findings: Among adult managed-care-eligible MassHealth members with at least one acute hospitalization, 8.7% were high-frequency hospital users; they contributed 30.2% of hospitalizations and 69.4% of readmissions. High-frequency users were more often male (77.1% vs. 50.0%; P < 0.001) and sicker (mean medical morbidity score was 3.3 vs. 1.9; P < 0.001) than others. They also had significant social risks: 33.1% with housing problems, 44.1% disabled, 83.2% with serious mental illness, and 77.1% with substance abuse disorder (vs. 22.0%, 27.3%, 60.2%, and 50.0%, respectively, for other hospital users [all P values <0.001]). Fully 50.7% of hospitalizations for high-frequency users led to 30-day readmissions (vs. 9.7%), contributing 72.0% of the variance in 30-day readmission, and substantially affecting judgments about the relative performance of accountable care organizations.

Effects of the Affordable Care Act Medicaid Expansion on the Distribution of New General Internists Across States
José Escarce et al.
Medical Care, July 2021, Pages 653-660

Research Design: Using 2009–2019 data from the American Medical Association Physician Masterfile and information on states’ Medicaid expansion status, we estimated conditional logit models to compare where new physicians located during the 6 years following implementation of the expansion to where they located during the 5 years preceding implementation.

Results: Thirty-three states and the District of Columbia expanded Medicaid by the end of the study period. Compared with preexpansion patterns, we found that physicians in one specialty group—general internal medicine—were increasingly likely to locate in expansion states with time after the expansion. The Medicaid expansion influenced the practice location choices of men and international medical graduates in general internal medicine; women and United States medical graduates did not alter their preexpansion location patterns. Simulations estimated that, between 2014 and 2019, nonexpansion states lost 310 general internists (95% confidence interval, 156–464) to expansion states.

The Spillover Effects of Capacity Pooling in Hospitals
Jong Myeong Lim, Hummy Song & Julius Yang
University of Pennsylvania Working Paper, March 2021


Off-service placement is a common capacity pooling strategy that hospitals utilize to address mismatches in supply and demand that arise from the day-to-day variation in patient demand. This strategy involves placing patients in a bed in a unit that is designated for another specialty service. Building on prior work that documents the negative first-order effects of off-service placement on patients who are placed off service themselves, we quantify the spillover effects of this practice on patients who are actually placed on service. Using an instrumental variables approach, we find that patients placed on service experience a substantial spillover effect of off-service placement, manifesting as a longer length of stay, a higher likelihood of hospital readmission, and a higher likelihood of clinical trigger activation. Our analyses show it is not only the average level of off-service placement, but also the volatility of off-service placement, that contributes to this spillover effect. Through a series of counterfactual analyses, we propose alternate routing policies that could meaningfully improve outcomes around the efficiency and quality of care in the inpatient setting.

Paying off the Competition: Market Power and Innovation Incentives
Xuelin Li, Andrew Lo & Richard Thakor
NBER Working Paper, June 2021


How does a firm’s market power in existing products affect its incentives to innovate? We explore this fundamental question using granular project-level and firm-level data from the pharmaceutical industry, focusing on a particular mechanism through which incumbent firms maintain their market power: “reverse payment” or “pay-for-delay” agreements to delay the market entry of competitors. We first show that when firms are unfettered in their use of “pay-for-delay” agreements, they reduce their innovation activities in response to the potential entry of direct competitors. We then examine a legal ruling that subjected these agreements to antitrust litigation, thereby reducing the incentive to enter them. After the ruling, incumbent firms increased their net innovation activities in response to competitive entry. These effects center on firms with products that are more directly affected by competition. However, at the product therapeutic area level, we find a reduction in innovation by new entrants after the ruling in response to increased competition. Overall, these results are consistent with firms having reduced incentives to innovate when they are able to maintain their market power, highlighting a specific channel through which this occurs.


from the


A weekly newsletter with free essays from past issues of National Affairs and The Public Interest that shed light on the week's pressing issues.


to your National Affairs subscriber account.

Already a subscriber? Activate your account.


Unlimited access to intelligent essays on the nation’s affairs.

Subscribe to National Affairs.