• Health care spending variation in Medicare and private coverage ain’t the same thing

    There are lots and lots of people who believe that increased health care spending must involve higher quality. Otherwise, we’d be forced to admit that we could spend a lot less and suffer no real consequences. A new viewpoint in JAMA summarizes a recent report from the IOM that begs to differ (emphasis mine):

    Recognizing that unexplained geographic variation in medical expenditures is both persistent and possibly a sign of inefficiency, members of Congress asked the Institute of Medicine (IOM) to explore geographic variation and address its policy implications. We chaired the resulting IOM committee, which recently issued its final report. The committee scrutinized the magnitude and implications of geographic variation in Medicare expenditures and examined the variation in health expenditures for patients whose care was financed from other sources. In its analyses of geographic variation in overall area-specific medical expenditures, the committee estimated spending by private insurers, Medicaid, the uninsured, and other out-of-pocket spending in an effort to measure the per-capita dollars spent on health care from all payment sources. In short, the committee sought to compare the total dollars spent on health care in different areas.

    The committee found that geographic variation in both Medicare and private expenditures was a real phenomenon and that high-cost areas in 1 year had high costs in other years. Areas with high costs in the treatment of 1 condition were somewhat more likely to have high costs in treating other conditions, but higher expenditures were not associated with better outcomes or higher quality care, either overall or when viewed from the perspectives of individual health conditions. Importantly, Medicare spending in an area and input-price-adjusted spending by private insurers were poorly correlated (with correlation coefficients of less than 0.1).

    It seems that most of the variation in per beneficiary spending is in “post-acute care”. This includes home health services, nursing homes, rehab facilities, long-term care hospitals, and hospice care. In fact, about three quarters of variation is accounted for by these services, which account for about 13% of all Medicare spending. The rest of the variation comes from inpatient services. Variation in outpatient services was negligible.

    Here’s something else that’s interesting (emphasis mine):

    These findings from Medicare do not generalize to private health insurance, for 2 principal reasons. First, post–acute care services (the main source of variation in Medicare spending) are seldom used by the younger privately insured population. Second, whereas Medicare sets a take-it-or-leave-it price after adjusting for wage levels, private insurers individually negotiate prices with physicians, hospitals, and other health care institutions and practitioners. The variation in negotiated prices presumably reflects variation in the local market power of both the insurer and the health care entity or clinician, as well as any variation across markets in efficiency among clinicians, hospitals, and other health care institutions. These 2 differences may help explain the poor correlation between Medicare and private insurer costs.

    Whereas price variation explains almost none of the overall variation in Medicare expenditures (after adjusting for wage variation), price variation is responsible for an estimated 70% of the total geographic variation in spending among privately insured persons. Variation in wage levels and variation in the quantity of services delivered are almost equally responsible for the remaining estimated 30% of spending variation. Thus, although most of the variation in per-beneficiary Medicare spending across HRRs is in the quantity of services, especially post–acute care services, the prices insurers pay in different areas—not the quantity of services—are responsible for most of the variation in private spending. These results are consistent with an important finding concerning Medicare: there is relatively little variation in either Medicare or among the privately insured in the quantity of services predominantly used by those younger than 65 years, such as outpatient visits and procedures. This helps explain why there is almost no correlation between Medicare spending and commercial spending in an area.

    Bottom line: Medicare spending in an area tells you almost nothing about private spending in an area, and vice versa. Moreover, what we might do to reduce one may have little effect on the other. Medicare has a volume problem, while private insurance has more of a price problem. We’ve got our work cut out for us.


    • Timely. I posted yesterday on another just released study also highlighting the importance of postacute services in considering cost variation:


    • This comment is a little off topic, but not much. We are about to see a revolution in post-acute care: wireless monitoring of patients. Developed on the battlefield by the military for wounded soldiers, wireless monitoring is now being developed for use in the civilian sector (with the initial emphasis in capnography). Hospitals and insurers believe they will be extremely effective in post-acute care, significantly reducing hospital re-admissions for example, and are working with the developers of the technology. I’m both hopeful (because it’s one of my projects) and concerned (because many of the physicians I work with – such as cardiologists – fear that the capacity to use the information effectively will be outpaced by the technology). We often hear that government can’t do anything right. Yet, the revolution in the use of computers was attributable to our space program, and the coming revolution in the use of wireless technology in medicine is attributable to our military. One paradox, though, is that the technologies, computers and wireless monitoring, were first developed and used in the military (public) sector but then implemented in the civilian sector first before being implemented in the public sector (e.g., Medicare). Why is that? I have my theory, and it’s not that government can’t do anything right.

      • A purely anecdotal observation, but the seniors in my family who have post-acute care are going to rehab to improve mobility. Given that they are both well into their 80s with other issues, I’d think they will never, despite the best care, be as mobile as they wish to be.

        For them, wireless monitoring won’t do much to reduce costs.

    • It’s not altogether surprising that higher per-beneficiary Medicare spending would appear to be driven by use of post-acute services. We’ve known for a while that a small % of Medicare beneficiaries account for a very large % of spending each year – they’re the ones who are really sick, often with co-morbidities, who will have frequent hospitalizations and lots of referrals to post-acute providers. Those older adults diagnosed with cancer that are excluded from clinical trials, the ones TIE wrote about recently, they’re going to use a lot of services.

      Nothing new here, but worth remembering that hospitals are protecting their margins when they refer patients for rehab, home care, hospice or other post-acute services.

      Still, as I read the Dartmouth findings and conclusions, regional variation in Medicare spending is strongly influenced by the practice of MDs in some HRRs to hospitalize patients for medical treatment or surgical procedures much more frequently than their colleagues in other areas do for the same diagnoses. Does this study find or conclude something different? I also wonder about ICU use by patients during the last year or 6 months of life – % doesn’t need to be real high to drive spending up.