• Reducing uninsurance may benefit everyone, not just the uninsured

    As we once again begin to debate the limits of personal responsibility, and how much one of us needs to care for another, it’s worth thinking about the following. There’s little question that people without health insurance may not have optimal access to the health care system. But some have theorized that higher rates of uninsurance in an area may have a “spillover” effect that impacts even those with insurance:

    A high community uninsurance rate may generate adverse spillover effects on access to and quality of care among the insured through several mechanisms. One possibility is that physicians practicing in a community with a high uninsurance rate curtail their provision of unprofitable services and shorten hours of service, which could affect access and quality for the insured and uninsured alike. Another potential mechanism results from the fact that mainstream providers—particularly private physicians—often both treat insured patients and provide charity (free or discounted) care to the uninsured. Physicians may find it difficult to provide very different levels of care to their insured and uninsured patients, and instead may tend to provide a similar level of care to all patients. Thus physicians may treat all patients in accordance with the average or modal insurance coverage in an area (or in their practice).  Because the uninsured can afford—and hence demand—a lower quantity and quality of care than the insured, physicians who practice in communities with a high uninsurance rate, and who treat numerous uninsured patients as a result, may provide a lower level of care to all their patients than physicians who practice in communities with a low uninsurance rate.

    A recent study in Medical Care looked at just this:

    Background: Previous research suggests, but does not definitively establish, that a high level of uninsurance in a community may negatively affect access to and quality of health care for insured persons.

    Objective: To assess the effect of the level of uninsurance in a community on access to and satisfaction with care—an important dimension of quality—among insured persons.

    Research Design: The 1996 to 2006 Medical Expenditure Panel Survey Household Component data linked to data from the Current Population Survey, Area Resource File, and the InterStudy Competitive Edge. Analyses include 86,928 insured adult respondents living in approximately 200 large metropolitan areas.

    Main Outcome Measures: Measures of whether an individual had a usual source of care, had any delay/difficulty obtaining needed care, used office-based services, used prescription drug services, and used any medical services, and measures of satisfaction with care.

    What did they find?

    For every 10% increase in the level of uninsurance in a community:

    • the chance that a person with private insurance had a usual source of care went down by 6.2%
    • the chance that an person with private insurance had difficulty receiving needed care went up by 7.7%
    • the chance that an person with private insurance reported being satisfied with their usual source of care went down 1.9%
    • the chance that a Medicare enrollee had problems receiving care, or didn’t get it at all, went up 1.7%
    • the chance that a Medicare enrollee rated health care quality as high went down 6.8%
    The bottom line? Increasing the number of people with insurance might not just benefit the uninsured, but also those who already have insurance.
    • A problem in our community is that some practices treat charity and/or Medicaid and other practices don’t do so at all. Aside from the ethics of this, which I find troubling, it has created a problem for those practices that provide care to the uninsured, charitable or Medicaid patients — paying patients don’t want to share a waiting room with these people. Since they have the option to go elsewhere they do. Has anybody looked at how this impacts a community, the providers, etc?

      And a question regarding the survey results . . . has anybody looked at what patient satisfaction means? Isn’t it true that part of our medical economic problem is with unreasonable expectations? Who determined what needed care included . . . the patient or the MD?

    • Thanks so much for posting this paper–as a provider who routinely has to turn away insured patients from our specialty practice in order to take care of the uninsured, it is nice to know that this experience is not unique. Also, it is a great point to make with people that think they “shouldn’t have” to help pay for the uninsured. They not only have to pay more because of cost shifting, but they may lose out on getting the care they need as sicker uninsured patients fill up the system. TIE is great!!!

    • I would be curious to see how this reflects in an area with a larger Amish population.

      My father is a solo practice PCP and a decent percentage (I would guess around 40%) of his patient population is Amish. Through the religious exemption, they will not have to participate in the ACA at all since they do not take any type of government handout. His elderly Amish patients are not on Medicare, for example. With this type of population, he and the local Critical Access hospital work with a population who will never have health insurance.

      An area with a higher Amish population could be a good control group for a longitudinal study like this to see how care is affected before and after ACA implementation since they may be the only group of Americans who’s insurance is guaranteed not to change.

    • I don’t dispute the idea that high rates of uninsurance may affect people who have insurance, but I do wonder if some of the respondents who had insurance but had recently between uninsured (more likely in areas with high rates of insurance ? ?) might have confused problems they had when uninsured with problems they had while insured.

      The limited description of the study I had access to doesn’t describe the exact questions people were asked, but even if they were as precise as possible people sometimes have problems remembering exactly when something occurred.

    • “Post hoc ergo proper hoc.”

      10 to one you could pick average income, average level of education, average number of vintage wine enthusiasts, and find a statistical connection with all of the above that’s just as robust, if not more so, than the connections cited between them and the percentage of uninsured.

      I’ll go out on a limb and suggest that the qualify of care, level of patient satisfaction, level of provider availability, etc for insured and uninsured alike are all highly correlated with level of participation in polo leagues, golf courses that host PGA events, and subscriptions to “Wine Enthusiast” magazine.

      Of course there’s less medical care available, and it’s not as good in area’s where there’s less money to pay for it. Does anyone really believe that the statistical association would hold if everyone in BelAir decided to self insure?

      • Your limb would likely break. These aren’t amateurs. They control for a lot. From the paper:

        Explanatory Variables

        The key explanatory variable in our analyses was the community uninsurance rate in the metro area, defined as the number of individuals less than age 65 who reported being uninsured divided by the population of individuals under age 65 in the metropolitan statistical area. We measured this using a 3-year moving average based on CPS data from 1995 through 2007.

        Individual-level covariates in our analyses included age, sex, age-sex interactions, and indicator variables for family income, educational attainment, nativity, language of interview, marital status, and family size. In addition, the models for insured working-age subjects included an indicator for whether or not the person was enrolled in an HMO, and the models for subjects with Medicare coverage included an indicator for dual Medicaid eligibility. Additional individual-level covariates included health status measures spanning four domains: (1) functional, cognitive, and social limitations; (2) vision or hearing problems; (3) self-rated general and mental health; and (4) chronic conditions.

        Contextual covariates included HMO penetration, the percentage of the population with income less than the federal poverty level (FPL) and with income between 1 and 2 times the federal poverty level, and the physician-to-population ratio for general internists, family physicians, and general practitioners. Finally, all models included indicators for the size of the metro area and year of the MEPS data.

    • Until someone runs a statistical regression that compares all of the above metrics to the average price per square foot of private housing I suppose that will have to be relegated to the realm of idle speculation. Having said that, though, it’s worth asking to what extend variations in the availability and quality of care persist under single payer regimes. Since it’s clearly not insured status that’s driving the variances once you’ve eliminated lack of coverage as a variable.

      I’m not personally convinced that normalizing on the basis of high level aggregates fully accounts for all of the characteristics that distinguish both the inhabitants and the environments of say, Mass-General’s catchment area (Newton, Brookline etc) to those of Boston University’s (Dorchester, Roxbury, Mattapan) – but I suppose that well intentioned people can disagree on that point.

    • BTW – the argument that I’m presenting in the context of this study doesn’t seem to differ significantly from Austin’s critique of claims based on statistical associations between Medicaid and negative outcomes that largely persist even when you attempt to control for the differences in Medicaid vs non-Medicaid populations.

      “However, that does not imply that the populations on Medicaid are equivalent to those not on Medicaid, even controlling for observable factors. We should think more deeply about why people who are otherwise nearly identical by all observable factors are either on or off Medicaid. Why, for example, would someone not be on Medicaid one month and then enroll the following even if they were eligible both months? Or, if it is a small change in income that confers eligibility, why did that income decline or rise? Do the observable factors explain everything? Many studies, which I’ve cited (see also the links above), show us they do not.”