Health Affairs is a peer-reviewed journal, which is why it was surprising to see it publish a recent article on the Massachusetts health reform, by Long et al [gated, but with abstract]. Based on telephone surveys, the authors declare that RomneyCare “continued to fare well in 2010.” This is an important finding, as the authors consider RomneyCare “the template for the federal Affordable Care Act of 2010.”
Unfortunately, in several cases the authors fail to inform readers that their results are contradicted by other, possibly more reliable, sources of information. They also neglect to put some of their results in proper context. Some examples:
- Failing to mention that although the nonelderly adults in the telephone survey samples reported a drop in emergency department (ED) use from 2006 to 2010, data from other reputable sources suggest that total ED visits have risen.
- Failing to mention that the “strong and sustained gains in the share of nonelderly adults in Massachusetts who reported their health as very good or excellent” are similar to the gains reported by all American adults.
- Misrepresenting the historical record with the claim that the “Massachusetts 2006 health reform initiative did not tackle the high cost of health care in the state.” Readily available sources clearly show that reform proponents expected it to reduce health care costs.
- Asserting that “access to health care in the community is better than it was in 2006,” without reporting on evidence that contradicts this conclusion.
- Inappropriately limiting the definition of “affordability” to out-of-pocket expenses, while ignoring higher premiums, fees, and taxes.
- Concluding that the survey evidence “is suggestive of important improvements in the effectiveness of the delivery of health care in the state” and implying that this is due to RomneyCare despite noting, three times on one page, that “our data did not allow us to isolate the impact of reform from that of changes in other factors during the study period.”
Emergency Department Visits
The telephone survey data suggests that a smaller fraction of nonelderly adults made emergency department visits both in general and for preventable conditions in 2010 than in 2006. However, the Massachusetts Division of Health Care Policy and Finance reported that outpatient emergency department visits increased 9 percent from 2004 to 2008, and that while visits for emergency conditions were stable, preventable/avoidable visits increased by 13 percent. Within the preventable/avoidable visit category, primary care treatable visits increased by 13 percent.
Chen et al. (NEJM 2011) concluded that “Massachusetts’ health care reform law has thus far neither increased nor decreased ED utilization relative to that in other states [Vermont and New Hampshire].” They used data on people under 65.
Perhaps emergency department visits by nonelderly adults did fall and the entire increase in total visits was a result of an increasing frequency of visits by children and elderly adults. The problem is that a reader of this article has no idea that the authors’ results are in conflict with those of other researchers, or why the telephone surveys in this paper should be considered a more reliable guide to reality than the data used by Chen and the Massachusetts state government.
By the way, Joshua Archambault of Massachusetts’ Pioneer Institute notes that people are unlikely to tell a telephone interviewer that they went to the emergency room unnecessarily.
So you lie, buy a little time, and I go along
What else can I do, maybe it’s wrong
Gains in Self-Reported Health.
The article relies on gains in self-reported health to support its contention that RomneyCare improved health. The authors write that “[we] found strong and sustained gains in the share of nonelderly adults in Massachusetts who reported their health as very good or excellent, with an increase from 59.7 percent in 2006 to 64.9 percent in 2010 (data not shown).”
What the authors did not do is inform the reader that these “strong and sustained gains” may simply be part of a national trend. In the 2006 National Health Interview Survey, about 61 percent of all US adults reported being in very good or excellent health. By 2010, almost 66 percent of adults reported being in good or excellent health. The National Health Interview Survey included elderly adults as well as nonelderly adults. Elderly adults are usually in poorer health than nonelderly adults, and one would therefore expect that the nonelderly adults surveyed in the article would report better health on average than participants in the National Health Interview Survey unless, of course, Massachusetts health care has systematically been below average.
It is also possible that RomneyCare exerted a placebo effect on self-reported health status. As Michael Cannon reported, when people in Oregon were given Medicaid coverage, “two-thirds of the improvement in self-reported health occurred almost immediately after enrollment, before any increases in medical consumption.”
Misrepresenting the Historical Record
The authors of this paper, along with a number of recent RomneyCare apologists, claim that the Massachusetts reform wasn’t really about reducing costs.
Yet, in an April 11, 2006, op-ed in The Wall Street Journal, then Governor Mitt Romney wrote that “Every uninsured citizen in Massachusetts will soon have affordable health insurance and the costs of health care will be reduced. And we will need no new taxes, no employer mandate and no government takeover to make this happen.” According to a January 18, 2006 article in the Boston Globe, a Families USA study said that health reform would “reduce pressure on health insurance premiums by at least $501 million a year.” Goals mentioned in the RomenyCare statute language include lowering or containing the growth in health care costs, reducing hospital administrative costs, and reducing the cost of insurance premiums.
Assuming that “Access” Has Improved
The article explains that RomneyCare was “expected to affect access to and use of care along two paths: by expanding access to health insurance and by creating a new standard that health plans must meet to count as coverage under the individual mandate.” In company with an unfortunate number of analysts, the authors implicitly assume that coverage and medical care are the same thing. Increases in “coverage” do not measure access to medical care because health plans can deny access to care by paying so little that no one will see their patients, restricting access to physician and hospitals, making people wait for care, and simply refusing to cover a given procedure.
Massachusetts increased coverage by a) mandating that people have health insurance while b) basically giving it away to roughly the bottom half of the income distribution. It also mandated enrollment in state run programs.
In all, the Massachusetts Division of Health Care Policy and Financing estimates that RomneyCare resulted in 388,196 newly insured people from 2006 to 2010, only 55,167 (14 percent) of whom paid the entire cost of their own policies. Yelowitz and Cannon(2010) remind readers that it is possible that the number of people with coverage is overstated. They note that responses to survey questions about coverage have fallen, possibly because respondents who say they are uninsured would be admitting that they are breaking the law.
In giving away coverage, Massachusetts encouraged the newly insured to treat health care as an almost free good. This increased demand at a time when the supply of medical care remained fairly constant. In fact, the state’s pressure on reimbursement negatively affected hospital margins, pressure that, over time, may reduce the supply of medical care. As the RAND Health Insurance Experiment showed, people with essentially free care increase utilization, but that utilization increase is not necessarily associated with better health.
As evidence that access has improved, the authors cite increases in visits to health care providers other than primary care physicians, hospitals, or emergency departments. They also cite the increased likelihood of “having a usual place to go for care” as evidence that access has improved. But the kind of access discussed in the article does not matter if there is a mismatch between the services that people want and the services that are offered. The authors note that there was a drop in the share of adults reporting a general doctor visit and suggest that this “may imply a shift in use toward other providers for some needs.”
In 2009, a Kaiser Family Foundation report on Massachusetts Community Health Centers by Ku et al. noted that the RomneyCare reform may have increased the demand for care at the centers and worsened the existing provider shortage. It mentions that the Community Health Centers responded by maintaining “months-long” waiting lists. In 2011, the Massachusetts Medical Society reported that “access to primary care physicians is becoming more restricted” and that there are longer waits for appointments in the four specialties it surveyed.
The authors report that their regression-adjusted estimate of the percent of nonelderly adults reporting a stay in the hospital fell slightly from 2006 to 2010. It is unclear how important this is. Among other things, the decline could have been due to statistical variation, the recession, managed care rationing, or individual choice. Nationally, the total number of hospital stays for nonelderly adults appears to have been stable for this period at roughly 39.4 million, the rate of hospitalizations per 10,000 people remained essentially the same from 1997 to 2009, and there is evidence that hospital admissions, especially for elective procedures, have fallen as a result of the recession.
Using an Inappropriate Definition of Affordability
Though the article maintains that health care has become more “affordable,” it provides data only for out-of-pocket spending.
For all the hoopla over Massachusetts’ success in increasing coverage, there has been surprisingly little study of RomneyCare’s total costs. Using insurance to pay for routine medical care is an expensive way to purchase health care because insurer overhead must be added to the cost of care. People who buy health insurance understand that lower out-of-pocket costs are associated with higher premiums. The authors claim that RomneyCare improved access by requiring third party payment for a “comprehensive set of services” and by limiting out-of-pocket payments. But free people buying their own health insurance generally choose to reduce their total costs by trading higher out-of-pocket costs for lower premium, higher deductible policies. In addition, they often chose more limited coverage than that required by RomneyCare’s “comprehensive set of services.”
In the real world, where real people work hard to pay real money in taxes to support government programs, a serious assessment of the success or failure of an initiative requires demonstrating that its benefits exceed its total costs.
In 2011, Tuerck et al. of the Beacon Hill Institute at Suffolk University estimated the total cost of RomneyCare at about $22,000 for each newly insured person. Overall, the Division of Health Care Finance and Policy reports that spending per privately insured Massachusetts resident grew by 6 percent from 2007 to 2008 and by 10 percent from 2008 to 2009. These rates of growth were substantially higher than the increase in per capita national personal health expenditures (4.9 percent and 4.6 percent, respectively).
The Beacon Hill report suggests that private payers have borne slightly more than half of RomneyCare’s costs and that the state has managed to shift most of the rest of it to the federal government. This level of spending is obviously unsustainable. Alan G. Raymond, in a 2011 report produced for the Blue Cross Blue Shield Foundation of Massachusetts, warns that “moderating future growth in health care spending is far more difficult than achieving nearly universal coverage, but without cost control, coverage expansions are unsustainable.
The Massachusetts Division of Health Care Policy and Finance apparently agrees. It hired RAND Corporation to develop a “comprehensive menu” of cost containment options. The 2009 RAND report suggested 21 possibilities. Virtually all of them involve new price controls, new ways to reduce access to care, or newly fashionable nostrums dreamed up by wonks eager to tell health patients and physicians how to go about their business.
On the Survey Evidence
For the record, the telephone surveys that generated the data for the article all had response rates under 50 percent. Though the authors undoubtedly considered how the people who did respond differed from those who did not, no mention of this appears in the article. Rather than detail much in the way of methodology, the article directs readers to several related reports.
The paper does report that the authors pool all five years of data and “test for differences in the outcomes relative to 2006â€³ and, for each of the prior years for 2007 to 2010. They also “obtain regression-adjusted estimates for each year using the parameter estimates from the regression models to predict the outcomes that the individuals in the 2010 sample would have had if they had been observed in each of the preceding study years.”
One of the reports says that 7 to 11 percent of the sample refused to respond to questions on family income. Those data were therefore “imputed” using unspecified “hot decking procedures.” Hot deck imputation usually means using “similar” responding units to arrive at a reasonable income figure. Because the wrong income categories were assigned to the federal poverty level in the 2010 survey, adults in that survey were assigned to the “correct” income category based on the income category people with similar age, sex, race, education, family type and family size occupied in the 2009 Massachusetts sample for the American Community Survey.