Last week a study by Harvard University highlighted new results from Oregon’s 2008 randomized Medicaid expansion. That year the state utilized a finite amount of funding for a coverage increase in the joint-federal insurance program by conducting a lottery for enrollment. Consequently, health care researchers have had a rare opportunity to analysis the subsequent outcomes, sometimes referred to as the Oregon Health Insurance Experiment (OHIE), in something as close to a massive randomized control trial as one could get in this context. The most recent findings showed that enrollees who had gained coverage in the lottery visited the emergency department (ED) of hospitals more often — specifically they observed a 40 percent increase during the time period studied.
There results were, on the whole, perhaps surprising. Some supporters of the Medicaid expansion in the new health care law argued, like President Obama, that increasing coverage for low-income Americans would reduce the incidence of emergency department visits. Furthermore, that such a reduction might contribute to lowering overall costs. In that sense the latest research from OHIE is a disappointment for those advocates. For other, less supportive, commentators the results were another reason to believe that Medicaid represents a ‘humanitarian disaster.’ Insomuch as research derived from Oregon’s experiment has largely been used to reaffirm prior beliefs on the ideal structure of our national health care system, these new details seem to fit the bill. However, a more honest and helpful interpretation of the most recent Oregon Medicaid data are also readily available.
First, there are some caveats to this study. As Harold Pollack writes in his excellent response, it is “not entirely clear how these experimental results can be generalized,” that the people who “signed up for the Medicaid waiting list and pursued coverage if they won the lottery are a relatively high-need group.” Moreover, the time period studied was relatively short-term — 18 months after the lottery, and only in Portland area hospitals. So while the research may adhere to a high degree of internal validity, it’s applicability outside of that context is much more limited. Extrapolating this to some firm expectation of similar results in states expanding Medicaid under the new law would necessarily be flawed at best. Even within the state of Oregon the data may not be as relevant, given that they have since pursued intrastate reforms that have reduced ED visits.
That being said, for health care experts these results were not actually counter-intuitive. As Adriana McIntyre wrote there are several reasons why low-SES (socioeconomic status) individuals would choose the emergency department at a higher rate after obtaining coverage; she mentions a Health Affairs paper showed that such individuals associated a high time cost to primary care visits and believed that the ED provided superior care. Moreover, as Austin Frakt noted, individuals “may rationally opt to visit the emergency department (ED) when a less intensive, ambulatory setting would suffice for many reasons of convenience, culture, and/or ignorance.” These varying reasons could be useful points of analysis with regard to our health care system, but whether they justify the existence of Medicaid is not an empirical judgement.
Another important factor to keep in mind is that decisions to visit the ED after obtaining coverage are probably embedded within a practical framework of restricted opportunity. A RAND corporation research report from last year (h/t Seth Trueger) found that respondents to Community Tracking Survey reported the following efforts when seeking care for acute (non-accident/injury) health problems:
As you can see from the figure above a slight majority (58 percent) had not contacted a primary care professional before their ED visit. Of that number most (84 percent) had attempted to contact “a doctor or other health professional,” while at the same time most (86 percent) could not have gone elsewhere. Against the backdrop of privately-insured individuals following similar patterns of overuse, perhaps it shouldn’t be surprising that coverage that increases access leads to increased use.
These results seem to indicate that, even outside the context of socioeconomic needs, the general picture of ED visits illuminate a more pervasive issue. Indeed, as Frakt sees it, overuse “reflects the broader problem that the health system is not very responsive to consumer demand or sensitive to all types of consumers […] it’s not consumers making “bad” choices, but the system offering poor ones.” That system is the way in which we actually deliver health care. There are some parts of the Affordable Care Act that seek to change that aspect (look up ACOs), but the new law is primarily an effort to change the way we provide health insurance. Changing what we don’t like about our delivery system is going to require even more work and is a reminder that health care reform is far from over.