By Julie Appleby
Kaiser Health News, July 15, 2011
What is Medigap and why do people buy it?
Unlike most job-based health insurance, traditional Medicare does not include “catastrophic” coverage, an annual maximum upper limit on the amount beneficiaries could pay. So enrollees can be liable for thousands of dollars each year, including: $1,132 per-episode deductible for hospital admissions; hundreds of dollars in daily charges for hospital stays of longer than 60 days; a $162-a-year deductible for doctor care, plus 20 percent of charges for office visits or equipment like wheelchairs.
Ten standardized types of supplemental plans offered by private insurers – including AARP’s UnitedHealthcare policies – cover all or most of such deductibles and copayments. Some employers also pay all or part of such costs for their retirees.
What changes are under consideration?
It is not clear exactly what’s on the table in the negotiations between congressional leaders and the White House. But the charts released show that one such proposal under consideration would bar insurers from offering supplemental policies unless the policies came with an annual deductible. People who didn’t want a deductible could pay $530 a year in additional premium to ensure that they won’t be hit with costs before their coverage kicks in.
What about people who don’t have a Medigap plan?
Only about 10 percent of seniors don’t have some sort of supplemental coverage. Some people have military/VA benefits, others are in Medicaid, and some have coverage through Medicare Advantage plans, which are insurance policies offered by private insurers as an alternative to traditional Medicare.
Would changing supplemental coverage save money?
Some economists and policy experts say that supplemental coverage insulates beneficiaries from medical costs, driving up demand for unnecessary care. A study done for MedPAC in 2009 found that beneficiaries with supplemental insurance used more care and cost the program more money. The increased spending wasn’t for emergency hospitalizations, but for other services such as elective hospital admissions, preventive care, doctor office visits and some types of tests.
What else do people say about the idea?
Advocacy groups like the Medicare Rights Center oppose restricting Medigap plans, saying it would simply shift more costs from the government to elderly and low-income people who can least afford it. “Some in government feel people in Medicare don’t have enough ‘skin in the game,’” says Ilene Stein, federal policy director for the center. In fact, she says, people on Medicare already pay 15 percent of their incomes for health care, well above the level paid by non-Medicare households. While the proposals would cap maximum annual spending per enrollee to $5,500 or $7,500, “that’s a lot of money for someone making $22,000,” the median household income for those on Medicare, she says.
Exploring the Effects of Secondary Coverage on Medicare Spending for the Elderly
Study by Direct Research, LLC
For the Medicare Payment Advisory Commission (MedPAC)
Our results show that secondary insurance has a substantial impact on Medicare spending, consistent with the prior literature in this area. After removing beneficiaries with any VA use and adjusting for differences in health status, income, education, and demographics, individuals with Medigap coverage had Medicare costs 33 percent higher than those with no secondary insurance. Other private secondary insurance was associated with smaller increases in spending. There was no statistically significant difference in Part A spending, but a large and statistically significant increase in Part B spending.
Out-of-pocket payment reduced spending largely by suppressing elective care (broadly defined) as opposed to emergency care. In particular:
* Emergency care (ambulance use, emergency room visits, emergency and urgent hospitalizations) appeared unaffected by the presence of secondary insurance.
* Elective admissions, preventive care, minor procedures and endoscopies were strongly affected by secondary insurance coverage, with substantially higher use among those with private secondary insurance.
It is not possible to use observational (non-experimental) data to prove beyond a doubt that a causal relationship exists between secondary insurance and spending. For several reasons, however, this analysis strongly suggests secondary insurance (reduced out-of-pocket costs) genuinely causes higher spending, and is not merely associated with secondary insurance due to some other factors affecting both insurance demand and health care use. The following factors suggest that this is a causal relationship. First, beneficiaries themselves report that out-of-pocket costs are a significant reason for delaying care. Nearly 20 percent of beneficiaries without secondary insurance reporting delaying care due to concerns over cost, versus 5 percent of beneficiaries with private secondary insurance. Thus, survey data provide direct evidence that out-of-pocket cost is a mechanism by which secondary insurance increases demand for care.
Second, there was a clear dose-response relationship between depth of insurance coverage and increased spending. Those with first-dollar or nearly first-dollar coverage had much higher spending than others, regardless of secondary insurance status.
Third, only the depth of coverage mattered, not the type of secondary insurance. When a flag for (nearly) first-dollar coverage was included in the regression, the individual types of secondary insurance were no longer statistically significant determinants of spending. Low out-of-pocket cost was a sufficient explanation for all of the observed increase in demand, regardless of the source of the secondary insurance.
This finding of a universal effect of first-dollar coverage regardless of insurance type weakens any alternative explanation based on the specifics of insurance ownership (described below). Ultimately, it did not matter whether beneficiaries chose to purchase coverage or not, or earned coverage as a retirement benefit or not. The only factor that mattered was whether or not their Part B care was free or nearly free. There are two generic counter-arguments that can be used to explain the results of a regression analysis as something other than a causal relationship. The first is omitted variables bias. This is the possibility that some unobserved factors are strongly correlated with insurance and are strong determinants of spending. This could be some unobserved difference in health status, or merely a systematic difference in beneficiaries taste for or preference for health care use. If such factors exist, then the apparent relationship between insurance and spending shown by the regression is merely proxying for these unobserved factors. The second counter-argument is self-selection, for the types of insurance that are individually purchased. If beneficiaries bought secondary coverage in anticipation of having higher spending, then the causality runs from spending causing insurance coverage, and not the other way around.
Unobserved health status differences cannot plausibly explain these results. Any hypothesized health status differen
ces would have to be highly selective and secretive. They would only affect the need for Part B services but not Part A, only require elective care but not emergency care, and would only affect those who have near-first-dollar coverage and not others. Such factors would also have to be undetectable both by physicians (reporting diagnoses used in risk adjustment) and by beneficiaries (in their own self-reported health and functional status). to be strongly correlated with ownership of secondary insurance. That combination is implausible enough that we can reasonably dismiss it from consideration.
Unobserved differences in taste and preference for health care are impossible to rule out as an alternative explanation of the results. It is possible that, on average, beneficiaries who ended up with nearly-complete secondary coverage, regardless of the source of that coverage, had developed a taste for higher levels of health care use prior to becoming eligible for Medicare. Whereas beneficiaries with the same class of coverage, but paying at least 5 percent of costs, did not. We could think of no obvious mechanism that would generate such a strong correlation across all types of secondary coverage. But tastes and preferences are idiosyncratic and unobservable, so there is no obvious data-driven way either to rule that out or to test it as an alternative.
Self-selection as an alternative hypothesis could only apply to individual purchase insurance, not to employer-sponsored coverage. If self-selection is offered as an alternative explanation, it has to be paired with some alternative explanation of higher costs for those with employer-sponsored coverage. Moreover, any self-selection of insurance based on observable factors – observed health status, income, education, or demographics – should largely be accounted for by the regression analysis. For example, any connection between good health and unwillingness to purchase secondary insurance should be captured by the presence of health status measures in the regression analysis.
Finally, we appeal to Occam’s Razor (a principle that generally recommends selecting the competing hypothesis that makes the fewest new assumptions, when the hypotheses are equal in other respects) to argue that the lack of copayment causes the higher spending by those with secondary insurance. On the one hand, one simple explanation – those who receive nearly free care use much more of it – provides a simple, universal explanation for the higher spending by beneficiaries with all types of private secondary insurance. On the other hand, alternative explanations are a hodgepodge of factors that only apply to some types of insurance (self-selection) and unobservable taste and preference factors that (through some unexplained mechanism) apply only to a subset of persons with secondary insurance (those with near-first-dollar coverage). Clearly, lack of copayment is the simpler explanation of what we have observed.
In summary, the evidence is reasonably clear that secondary insurance raises Medicare costs. After eliminating persons with VA use and adjusting for covariates (health, income, education, demographics), beneficiaries with secondary insurance use much more health care than those who have no secondary insurance. The effect is due solely to those with near-first-dollar coverage (defined here as paying less than 5 percent of Part B costs). Beneficiaries without such coverage, by contrast, appear no different from those with no secondary insurance. The differential impact by service type – more on Part B than Part A, more on elective care than emergency care – also suggests that the out-of-pocket cost causes the lower use of care. When asked, beneficiaries themselves say exactly that – those without secondary insurance are far more likely to report having delayed care due to cost. Taken together, this provides a coherent picture that out-of-pocket costs matter significantly to Medicare beneficiaries, and that eliminating those costs raises health care spending.
This analysis does not address whether the increased spending is desirable or undesirable, or whether reduced spending leads to poorer outcomes. That question — whether the value of additional care exceeds its cost – cannot be answered from the analysis of spending data alone, if it can be answered at all. Instead, this analysis merely shows that beneficiaries in fee-for-service Medicare will tend to use much more health care when each additional service is free (to them) than they would if they had to pay a significant portion of the cost of each additional service.
By Don McCanne, MD
Although this excerpt from the 44 page MedPAC report on secondary insurance for Medicare beneficiaries (Medigap, etc.) is more than most people want to read, it really is important because this report is being used to try to expand policies that would reduce the moral hazard of insurance. The theory is that people use too much health care if it is “for free” (no deductibles, copayments, or coinsurance), and that they would access only the care that they really needed if they had to pay for at least a portion of it. Is this theory valid?
Although many proposals to reduce the government component of Medicare spending would do so by transferring government costs to the beneficiaries, this proposal to require cost sharing in Medigap plans is quite different. Since Medigap plans are privately funded, the reduction in premiums due to increases in cost sharing would accrue to the beneficiaries and not to the government. The reduction in government spending occurs only indirectly by not having to pay the government’s share of that care which was not received because of the patients’ own concerns about out-of-pocket costs.
The authors of the MedPAC report do confirm that individuals without supplementary coverage (without Medigap) use less care and therefore cost less. Policymakers using this report seem to be extrapolating the conclusion that eliminating cost sharing creates an increased demand for care that patients can do without, even though the authors clearly state that “this analysis does not address whether the increased spending is desirable or undesirable.” Nevertheless, the authors do seem to demonstrate some bias as they dismiss alternative explanations.
A few observations are warranted:
* Above all, the greater amount of care obtained by those with first dollar coverage cannot be dismissed as being excessive or wasteful. Although predominantly elective, these included preventive services and services that may reduce symptoms and improve quality of life. Preventive services were used only half as often by those with no supplemental coverage compared to those with additional coverage. A health care system should provide affordable access to more than just emergency services.
* Only about 10 percent of Medicare beneficiaries have no supplemental coverage, and it was this group that used fewer services. The authors seem to dismiss the facts that these individuals may not be culturally attuned to freely using the health care system, and many would self-select to not purchase additional coverage. They may decide that premiums for the plans are not worth the cost, or they may simply not like doctors and want to avoid all of the health care that they can.
* The fact that those receiving supplemental plans from their employers used less care than those purchasing their own Medigap plans suggests that the former group was diluted with individuals who would not have purchased plans had their employers not provided them. Again, this suggests that individuals averse to health care and not wanting to pay for it should not be used as the standard for the proper level of care.
* An observation buried
in this report is that mortality was higher in those who had no supplemental coverage. Although the authors were not able to explain precisely why, they did state, “In summary, the importance of this finding comes down to a judgment. On the one hand, the finding of excess mortality is based on a single year’s experience, and is inconsistent with the other measures of population health status. On the other hand, it is not explained by demographic or other factors, and it is not obviously an artifact of the methodology used. Finally, excess mortality would be a reasonable outcome from deficiencies of care, for example, from the lack of preventive care for this population.”
* Whether or not increased utilization by those with supplemental coverage provided a medical benefit, the plans clearly provided financial security for users of health care. How could anyone consider that to not be of benefit?
* This report remains silent on the issue of the profound administrative waste in providing plans that supplement the traditional Medicare program. The savings achieved by rolling the benefits of the supplemental plans into Medicare would likely offset much of the cost of services declined because of perceptions of affordability. Of course, once we fix Medicare, we should provide it to everyone.
Blaming our high health care costs on the moral hazard of having free access to health care at the time of need has got to end. Several nations that spend half of what we do nevertheless provide comprehensive health care services with free access. The real moral hazard is not that people might use more health care that they don’t have to pay for at the time of services, the moral hazard we should be concerned about is that people might not get the care that they should have because of financial barriers to that care.
That turns the moral hazard argument upside down, but that’s where it belongs.
For more on cracks in the foundation of moral hazard, citing the work of John Nyman: