Research

Evidence Reversal in Medicine: Do Legal Standards of Care Affect Phase-out of Ineffective Practices?

Abstract: In medicine evidence reversals occur when current clinical practice is deemed ineffective or harmful, and evidence shifts in favor of an older standard of care or no treatment. Anecdotal evidence suggests that after an evidence reversal occurs the phase-out or "de-adoption" process for the procedure in question may take over a decade. Understanding barriers to de-adoption contributes to our knowledge of the determinants of productivity in the healthcare sector. Medical malpractice incentives may foster or discourage the implementation of new medical evidence or innovations that improve productivity. This paper investigates the role of a particular legal institution - definitions of the standard of care – in the de-adoption process for vertebroplasty following an evidence reversal in 2009. The legal literature hypothesizes that locality-based standards of care stifle progress by enabling laggards and punishing physicians who choose to practice at the cutting edge. Results indicate that overall, de-adoption of verteborplasty occurred rapidly in both locality and national standard states. Leveraging state-level variation in how medical malpractice is legally defined and within state variation in the practical application of these laws, results demonstrate that in rural areas, where locality rules should have a greater impact, de-adoption occurs more slowly, whereas de-adoption occurs more rapidly in urban areas of locality states. This study partially confirms predictions in the legal literature: while locality rules do matter for rural areas within states, as a whole they do not have a statistically significant effect on de-adoption.

Does Medicaid Affect Treatment Intensity and Mortality? Evidence from Inpatient Hospital Stays

Abstract: Does health insurance status affect the intensity of medical treatment? A substantial body of work shows that expanding coverage improves access to care on the extensive margin: more people consume care as a result of expansion. It remains unclear, however, whether insurance status affects the intensive margin: that is, the amount or type of care that a patient receives, conditional on receiving some care. In this paper, we analyze the impact of health insurance coverage on the intensity of treatment for nonelderly adults hospitalized for serious health conditions. The Medicaid expansion under the Affordable Care Act provides exogenous variation in coverage, while focusing on serious conditions for which hospitalization is essentially inevitable allows us to isolate the effects of coverage on the intensive margin. Data are from hospital discharge records for 316,000 inpatient stays in 13 states between 2011 and 2015.

We find that on average, the Medicaid expansion leads to a significant increase in Medicaid coverage that is partially offset by a decline in private coverage, leading to a small net reduction in uninsurance. Taken together, these changes in coverage have no statistically significant effect on average treatment intensity. However, this obscures the separate effects of coverage gains (transitions from no insurance to Medicaid) and crowding out (transitions from private insurance to Medicaid). In order to distinguish the effects of these different transitions, we repeat our analysis after dividing the sample into geographic areas with high and low rates of uninsurance prior to expansion. We find that in areas with high rates of uninsurance, the Medicaid expansion produces increases in Medicaid coverage without reductions in private coverage, and patients receive significantly fewer procedures. In-hospital mortality also increases significantly. In areas with low rates of uninsurance, the Medicaid expansion increases Medicaid coverage with fully offsetting reductions in private coverage, and there are no significant effects on treatment intensity or mortality. Thus, switching from uninsured to Medicaid appears to reduce treatment intensity, while switching from private coverage to Medicaid does not affect treatment intensity. Our results highlight the importance of allowing the effects of coverage expansion to vary depending on whether they induce gains in coverage (uninsured to Medicaid) or changes in the type of coverage (private to Medicaid, aka crowd-out).

Do Local Soda Taxes Affect Prices and Consumption? A Tale of Two Cities

Abstract: Many US cities have implemented or are considering soda taxes due in part to the growing

literature about soda’s negative health effects. As is the case with many public health interventions,

measured effects of local soda taxes vary by city and also vary among studies that analyze the same

city tax because estimates are sensitive to differences in methods and data sources. This study

estimates the effects of local soda taxes in Berkeley and Philadelphia using the same methods

and data in order to determine which measured differences can be attributed to local supplier and

consumer responses rather than methodology. Comparing the cases of Berkeley and Philadelphia

highlights which findings are constant across these very different cities and are likely to generalize

to other local soda taxes. Berkeley and Philadelphia make an interesting comparative case study

due to differences in motivation for the taxes (health vs revenue) and differences in demographics,

city size and the size of the tax base. The results from both cities indicate incomplete pass-through

rates that decline with beverage container size and moderate decreases in consumption of both

diet and regular soda. Price and consumption effects were larger in Philadelphia than in Berkeley.

Supply and demand elasticities can be separately identified using the tax as an instrument

Identifying Medical Reversals: An Introduction to a New Area of Study (NBER White Paper)