Healing Powers

Kevin Lewis

December 20, 2021

Diagnosing Physician Error: A Machine Learning Approach to Low-Value Health Care
Sendhil Mullainathan & Ziad Obermeyer
Quarterly Journal of Economics, forthcoming

How effective are physicians at diagnosing heart attacks? We study this question on a patient-by-patient basis by building a machine learning model of individual risk. We then contrast a physician’s decision to test a patient with the algorithm’s predicted risk; in cases of disagreement, we use data on health outcomes to judge whether the algorithm or the physician was right. These analyses reveal that physicians both overtest and undertest. Predictably low-risk patients are tested; and those tests yield little. At the same time, predictably high-risk patients go untested; and those patients go on to suffer adverse health events (including death). A natural experiment using shift-to-shift testing variation confirms these findings: additional testing would improve health (and reduce mortality), but only for patients flagged as high-risk by the algorithm. The simultaneous existence of over- and undertesting cannot easily be explained by incentives alone and suggests systematic errors. We provide suggestive evidence on the psychology underlying these errors. First, physicians use too simple a model of risk, suggesting bounded rationality. Second, they overweight salient information, Third, they overweight symptoms that are representative or stereotypical of heart attack. Together, these results suggest health care models for inefficiencies that arise from physician mistakes; and policies to address these may be very different from those aimed at addressing incentive problems. 

Did the Hospital Readmissions Reduction Program Reduce Readmissions? An Assessment of Prior Evidence and New Estimates
Engy Ziedan & Robert Kaestner
NBER Working Paper, December 2021

In this article, we provide a comprehensive, empirical assessment of the hypothesis that the Hospital Readmissions Reduction Program (HRRP) affected hospital readmissions. In doing so, we provide evidence as to the validity of prior empirical approaches used to evaluate the HRRP and we present results from a previously unused approach to study this research question — a regression-kink design. Results of our analysis document that the empirical approaches used in most prior research assessing the efficacy of the HRRP often lack internal validity. Therefore, results from these studies may not be informative about the causal consequences of the HRRP. Results from our regression-kink analysis, which we validate, suggest that the HRRP had little effect on hospital readmissions. This finding contrasts with the results of most prior studies, which report that the HRRP significantly reduced readmissions. Our finding is consistent with conceptual considerations related to the assumptions underlying HRRP penalty; in particular, the difficulty of identifying preventable readmissions, the highly imperfect risk adjustment that affects the penalty determination and the absence of proven tools to reduce readmissions. 

The Medicare Advantage Quality Bonus Program Has Not Improved Plan Quality
Adam Markovitz et al.
Health Affairs, December 2021, Pages 1918-1925

In 2012 Medicare introduced the quality bonus program, linking financial bonuses to commercial insurers’ quality performance in Medicare Advantage (MA). Despite large investments in the program, evidence of its effectiveness is limited. We analyzed insurance claims from the period 2009–2018 from the nation’s largest MA claims database for 3,753,117 MA beneficiaries (treatment group) and 4,025,179 commercial enrollees (control group). Using a difference-in-differences framework, we evaluated changes in performance on nine claims-based measures of quality in both groups before and after the start of the bonus program and with adjustment for differential pre-period trends. We observed no consistent differential improvement in quality for MA versus commercial enrollees under the quality bonus program. Program participation was associated with significant quality improvements among MA beneficiaries on four measures, significant declines on four other measures, and no significant change in overall quality performance (+0.6 percentage points). Together, these results suggest that the quality bonus program did not produce the intended improvement in overall quality performance of MA plans. 

The Impact of Expanding Public Health Insurance on Market-based Insurance Premiums: Roles of Risk-Pool Composition and Size
Qiang Liu, Ming Lei & Ting Zhu
Purdue University Working Paper, December 2021

The Medicaid expansion has a profound impact on Marketplace plan premiums. On the one hand, the expansion drove out the less healthy people from Marketplace plans, resulting in a healthier pool with a lower risk score for these plans. With this effect, we hypothesize that the Marketplace premium reduction due to the Medicaid expansion is less pronounced in more wealthy rating areas than in less wealthy ones. On the other hand, the Medicaid expansion may shrink the pool size of Marketplace plans, resulting in a riskier pool and a higher Marketplace plan premium. We hypothesize that the premium increase due to the Medicaid expansion is less pronounced in more populous rating areas than in less populous ones. Integrating data from Health Insurance Marketplace Public Use Files and the American Community Survey, we document empirical evidence consistent with these hypotheses by examining how the Medicaid expansion impact on Marketplace insurance plan premiums varies across rating areas with different demographics. We find that the Medicaid expansion decreased the Marketplace Silver plan premiums on average but did not change the Marketplace Bronze plan premiums on average. Our results offer important policy and managerial implications. 

Ambulance Taxis: The Impact of Regulation and Litigation on Health Care Fraud
Paul Eliason et al.
NBER Working Paper, November 2021

We study the relative effectiveness of administrative regulations, criminal enforcement, and civil lawsuits for combatting health care fraud. Between 2003 and 2017, Medicare spent $7.7 billion on 37.5 million regularly scheduled, non-emergency ambulance rides for patients traveling to and from dialysis facilities, with dozens of lawsuits alleging that Medicare reimbursed rides for patients who did not meet the requirements for receiving one. Using a novel data set and an identification strategy based on the staggered timing of regulations and lawsuits across the United States, we find that a regulation requiring prior authorization for ambulance reimbursements reduced spending much more than criminal and civil lawsuits did. Despite the sharp drop in both ambulance transports and the companies that provide them following prior authorization, patients’ health outcomes did not change, indicating that most rides were not medically necessary. Our results suggest that administrative actions have a much larger impact than targeted criminal enforcement, providing novel evidence that regulations may be more cost-effective than ex post ligation for preventing health care fraud. 

States’ Merger Review Authority Is Associated With States Challenging Hospital Mergers, But Prices Continue To Increase
Brent Fulton et al.
Health Affairs, December 2021, Pages 1836-1845

States can challenge proposed hospital mergers by using antitrust laws to prevent anticompetitive harms. This observational study examined additional state laws — principally charitable trust, nonprofit corporation, health and safety, and certificate-of-need laws — that can serve as complements and substitutes for antitrust laws by empowering states to be notified of, review, and challenge proposed hospital mergers through administrative processes. During the period 2010–19, 862 hospital mergers were proposed, but only forty-two (4.9 percent) were challenged by states, including thirty-five by states without federal involvement, of which twenty-five (71.4 percent) originated in the eight states with the most robust merger review authority. The twenty-five challenges resulted in two mergers being blocked; three being abandoned; and twenty being approved with conditions, including seven with competitive-impact conditions. Hospital market concentration and prices increased at similar rates in these eight states versus other states, potentially because most challenges allowed mergers to proceed with conditions that did not adequately address competitive concerns. Although these findings do not reveal an optimal state framework, elements of advanced state merger review authority may have the potential to improve poorly functioning hospital markets. 

Hospital Ownership and Managerial Influence over Treatment Decisions: The Case of Admissions from the Emergency Department
David Howard & Guy David
Emory University Working Paper, December 2021

A number of False Claims Act lawsuits have alleged that managers at for-profit hospitals pressured physicians, using feedback and manipulation of defaults, to admit patients unnecessarily from the emergency department. Using 100% hospital records from Florida for 2015-2017, we estimate that admission rates were 7 percentage points (20%) higher at for-profit hospitals. We additionally describe admission rate trends from 2007-2017 at two hospital systems that paid large penalties to the Department of Justice for unnecessary admissions. The timing of pre- and post-investigation changes in admission rates and our cross-sectional estimates are consistent with allegations that administrators influenced physicians’ admission decisions. 

Gender Differences in Medical Evaluations: Evidence from Randomly Assigned Doctors
Marika Cabral & Marcus Dillender
NBER Working Paper, December 2021

While a growing body of evidence documents large gender disparities in health care and related social insurance programs, little is known about what drives these disparities. We leverage administrative data and random assignment of doctors to patients inherent within the workers’ compensation insurance claim dispute resolution process to study the impact of gender match between doctors and patients on medical evaluations and subsequent social insurance benefits received. Compared to differences among their male patient counterparts, female patients randomly assigned a female doctor rather than a male doctor are 5.0% more likely to be evaluated as disabled and receive 8.5% more subsequent cash benefits on average. There is no analogous gender-match effect for male patients. The magnitude of these effects implies that having female doctors evaluate patients entirely offsets the observed gender gap in the likelihood of being evaluated as disabled when male doctors evaluate patients. We explore mechanisms through further analysis of the administrative data and complementary survey evidence. In addition, we present broader evidence on gender gaps in workers' compensation insurance and gender homophily in patients' selections of doctors in settings where patients have choice. Combining this evidence, we conduct policy counterfactuals illustrating how policies increasing gender diversity among doctors or increasing gender homophily in patient-doctor matches may impact gender gaps in evaluated disability. Our findings indicate that policies increasing the share of female patients evaluated by female doctors may substantially shrink gender gaps in medical evaluations and associated outcomes. 

In defense of Direct Care: Limiting access to military hospitals could worsen quality and safety
Cheryl Zogg et al.
Health Services Research, forthcoming

Data sources:
Adult MHS beneficiaries, aged 18–64?years, treated in MTFs (under a program known as Direct Care) were compared against (1) MHS beneficiaries treated in locally available civilian hospitals (under a program known as Purchased Care) and (2) similarly-aged adult civilian patients across the United States. MHS beneficiaries in Direct and Purchased Care were identified from fiscal-year 2016–2018 MHS inpatient claims. National inpatients were identified in the 2017 Nationwide Readmissions Database.

Principal findings:
A total of 502,252 MHS admissions from 37 MTFs and surrounding civilian hospitals were included (326,076 Direct Care, 179,176 Purchased Care). Nationwide, 9.34 million adult admissions from 2453 hospitals were included. On average, MHS beneficiaries treated in MTFs experienced better inpatient quality and improved patient safety compared with MHS beneficiaries treated in locally available civilian hospitals (e.g., summary observed-to-expected ratio for medical mortality: 0.98 vs. 1.03, p?<?0.001) and adult patients across the United States (0.98 vs. 1.02, p?<?0.001). Simulations of proposed changes resulted in consistently worse outcomes for MHS patients, whether reducing MTF access by 10%, 20%, or 50% nationwide; limiting MTF access to active-duty beneficiaries; or closing MTFs with the worst performance on patient safety (p?<?0.001 for overall quality indicators for each). 

The Impact of Provider Payments on Health Care Utilization: Evidence from Medicare and Medicaid
Marika Cabral, Colleen Carey & Sarah Miller
NBER Working Paper, November 2021

Provider payments are the key determinant of insurance generosity within many health insurance programs covering low-income populations. This paper analyzes the effects of a large, federally-mandated provider payment increase for primary care services provided to low-income elderly and disabled individuals. Drawing upon comprehensive administrative payment and utilization data, we leverage variation across beneficiaries and across providers in the policy-induced payment increase in difference-in-differences and triple differences research designs. The estimates indicate that the provider payment reform led to a 6.3% increase in the targeted services provided to eligible beneficiaries, indicating an implied payment elasticity of 1.3. Further, the provider payment reform decreased the fraction of low-income beneficiaries with no primary care visit in a year by 9%, completely closing the gap relative to higher-income beneficiaries with the same observable characteristics. Additionally, the results indicate that the payment reform caused an increase in established patient visits, with no increase in new patient visits. Heterogeneity analysis indicates that the payment increase led to an expansion of utilization for many subgroups, with somewhat larger effects among beneficiaries who are younger, are white, and live in areas with many primary care providers per capita.


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.