• ACOs aren’t getting all the Medicare data they need either

    A reader tipped me off to this letter from the National Association of ACOs.

    If the federal government is unwilling to reform patient consent for the purposes of an ACO’s ability to provide adequate care coordination we would recommend related claims data be provided ACOs as de-identified. This would at least allow ACOs to create better patient population profiles, enable ACOs to develop more effective programming and allow them to better manage costs or increase their accountability for costs.

    I haven’t chased down everything related to this, but it seems that ACOs aren’t receiving substance use disorder-related claims and don’t have other feasible ways to obtain the information they would convey.

    This is problematic because ACOs are responsible for clinical and cost outcomes for all patients, not just those without substance use diagnoses. And, ACO patients can obtain care from other providers, outside the ACO. So, how is an ACO supposed to plan and improve population care if it can’t obtain all the relevant information about their population?

    This is an important question. It is related to the one I’ve been asking about research data scrubbed of substance use-related claims. But, I should emphasize that these two issues, though related, are different. The National Association of ACOs is asking for de-identified Medicare data. Researchers often need identified Medicare and Medicaid data (e.g., to merge with that from other systems, like the VA). For this very reason, there are data security protocols in place for proper handling of research data.


    Comments closed
  • The waivers that could save Obamacare … if states want them to

    This is important:

    1332 waivers may also appeal to states with alternate Medicaid expansions, such as Arkansas and Iowa. So far, these so-called private option expansions, which enroll Medicaid-eligible individuals into private coverage, operate through Section 1115 waivers, which predate the ACA. But states may find the budget neutrality requirements of 1115 waivers to be overly restrictive. The ACA calls for a streamlining of the waiver process, whereby states can ask for 1115 and 1332 waivers in one application. As John McDonough wrote earlier this year, this combined waiver process could give states much more flexibility. For example, an 1115 waiver proposal that would not be independently budget-neutral could become acceptable in conjunction with a related 1332 waiver proposal. States will have greater ability to craft applications that meet the needs of their intended reforms.

    But next year’s Supreme Court term could have major ramifications on alternate expansion states and for 1332s more broadly. The innovation waivers offer states unparalleled flexibility in large part because they let them repurpose hundreds of millions of dollars in tax credits. In King v. Burwell, though, the Court will determine the availability of tax credits to residents of states that have not established exchanges. A ruling in favor of the plaintiffs would decimate the funding source for 1332-based reforms in those states.

    Thus, such a ruling would hamstring red states in particular. Policymakers seeking conservative, market-oriented changes to ACA at the state level would be stymied even before their reforms get off the ground. Only states that have established their own exchanges would have the freedom and funding to undertake broad 1332-based reforms.

    Go read the rest by Heather Howard and Galen Benshoof on the Health Affairs blog.


    Comments closed
  • An update on CMS’s data withholding

    I’ve received a lot of questions about CMS’s deletion of substance use disorder related claims from research Medicare and Medicaid files, which seems to have been done without public discussion.

    My fellow research colleagues from around the country are, rightfully, upset by this and want to know more. Why are the claims being removed? Who is benefiting or being protected by this? Which, specific claims are removed? Why was there no public process? Why isn’t there more information online? AUSTIN WHEN ARE YOU GOING TO GIVE US THE DETAILS!!!!???

    The fact is, nobody I have contacted (and I have contacted many) seems to know anything more than I’ve posted, or has been willing to tell me (so far) if they do know. I am working all channels and leads available to me. Even people I thought would know about this didn’t.

    I can only add two things, both meta:

    1. It’s very unusual for me to post a weedy/wonky/technical issue on this blog and not get answers, good ones. TIE, it seems, has deep and wide reach in the research and health policy communities. Based on that, my best assumption is that people who do know something about these data withholding issues know we’re seeking answers. Somebody reading these words somewhere knows something. Feel free to tell me! (Or, why can’t you?)
    2. The fact that this issue was not known to many researchers, and that details are this hard to tease out about a problem this important to research is why the community needs a space like this. I’m happy for TIE to serve as an alert system. I’m sounding the alarm here. Contact anyone you can think of who might know or know someone who knows about CMS data access decisions. Let me know what you find out. (If it must be off the record, so be it … but I gotta ask why that would be?)

    Lastly, watch this blog and/or Twitter. I will update as information becomes available.


    Comments closed
  • Health care social networks

    Here’s another paper I don’t have time to read, but might interest you. It caught my eye in part for its treatment of knowledge transfer and diffusion of innovation.

    Social Networks of Professionals in Health Care Organizations A Review“, by Stefano Tasselli (Medical Care Research and Review, 2014):

    In this article, we provide an overview of social network research in health care, with a focus on social interactions between professionals in organizations. We begin by introducing key concepts defining the social network approach, including network density, centrality, and brokerage. We then review past and current research on the antecedents of health care professionals’ social networks—including demographic attributes, professional groups, and organizational arrangements—and their consequences—including satisfaction at work, leadership, behaviors, knowledge transfer, diffusion of innovation, and performance. Finally, we examine future directions for social network research in health care, focusing on micro–macro linkages and network dynamics.


    Comments closed
  • The future of essential health benefits

    The concluding paragraph of my Upshot post today is a little out of date, it seems:

    The secretary of H.H.S. must revisit the essential health benefits regulations for 2016. Another debate about what they are and how much discretion should be left to states is not far off.

    That may have been correct when I drafted the piece, but toward the end of November, new, proposed regulations came out that clarified what HHS intends to do about essential health benefits. Tim Jost explained:

    [The current] benchmark plans will apparently continue in place through 2016. The proposed rule would allow states to choose a new base benchmark plans for 2017.  These plans will be based on 2014 plans, adjusted to ensure that they meet all EHB requirements. HHS proposes, therefore, to begin collecting data on 2014 plans for the purpose of identifying benchmark plans, including administrative data and descriptive information pertaining to covered benefits, treatment limitations, drugs, and exclusions.  Apparently, HHS does not intend to establish its own EHB requirements in the foreseeable future, but rather to continue to rely on the benchmark approach.

    Notice this is a proposed rule for 2017, so in theory there’s still opportunity for discussion and refinement (I guess). But it seems today’s standards will be in place through 2016, which isn’t quite what I wrote.

    This does not change the main point of my piece. With respect to essential health benefits, the ACA still doesn’t impose a costly, one-size-fits-all standard, and it won’t going forward.

    I thank Mitchell Stein for pointing out the discrepancy.


    Comments closed
  • Publicly financed family planning’s return on investment

    Return on Investment: A Fuller Assessment of the Benefits and Cost Savings of the US Publicly Funded Family Planning Program“, by Jennifer Frost, Adam Sonfield, Mia Zolna, Lawrence Finer (Milbank Quarterly, 2014):

    Policy Points: The US publicly supported family planning effort serves millions of women and men each year, and this analysis provides new estimates of its positive impact on a wide range of health outcomes and its net savings to the government. The public investment in family planning programs and providers not only helps women and couples avoid unintended pregnancy and abortion, but also helps many thousands avoid cervical cancer, HIV and other sexually transmitted infections, infertility, and preterm and low birth weight births. This investment resulted in net government savings of $13.6 billion in 2010, or $7.09 for every public dollar spent.

    A press release for the paper is here. We covered this area extensively in prior posts, for example herehere, and here.


    Comments closed
  • High hospital prices are not a clear signal of quality

    The following was originally posted on the AcademyHealth blog earlier this year.

    In 2010, Massachusetts Attorney General Martha Coakley published a report on health care cost drivers in the state. The report concluded that prices paid to hospitals in Massachusetts

    are not correlated to (1) quality of care, (2) the sickness of the population served or complexity of the services provided, (3) the extent to which a provider cares for a large portion of patients on Medicare or Medicaid, or (4) whether a provider is an academic teaching or research facility. Moreover, (5) price variations are not adequately explained by differences in hospital costs of delivering similar services at similar facilities.

    Price variations are correlated to market leverage as measured by the relative market position of the hospital or provider group compared with other hospitals or provider groups within a geographic region or within a group of academic medical centers.

    A similar theme has been struck in examinations of hospital markets in VirginiaOregon,  California. (More literature on hospital market concentration is summarized here and here.)

    A paper published in Health Affairs in February provides evidence that is in some ways consistent and in others inconsistent with the conclusion of the 2010 Massachusetts report. Chapin White, James Reschovsky, and Amelia Bond examined hospital claims data for the families of non-elderly autoworkers and retirees in 13 Midwestern markets, including 110 hospitals.

    In contrast to Coakley’s study, White et al. found that high-price hospitals were more likely to be major teaching hospital, to have a more private/less Medicare payer mix, to have a higher disproportionate share ratio (DSH, which measures the share of patients with low incomes, as would be the case for Medicaid enrollees), and a higher DSH revenue share.

    High-priced hospitals were not distinguishable from low-priced hospitals on most dimensions of quality, but they were statistically significantly of lower quality according to measures of surgical deaths and blood clots among surgical patients. They were statistically significantly of higher quality according to 30-day, risk-adjusted mortality rates for heart failure patients.

    However, high-priced hospitals were much more likely to be nationally ranked by U.S. News & World Report for ancer, cardiology, gynecology, or orthopedic care—often taken by patients as a measure of quality. However, according to the authors, those rankings “are based largely on hospitals’ reputations among physicians,” not on clinical indicators of high-quality care.*

    The findings are consistent with the Massachusetts report in an important respect: high-priced hospitals have greater market power, as illustrated in the chart below.

    white-hosp mkt share

    High-priced hospitals are also more likely to offer specialized services like neonatal intensive care, level 1 trauma care, heart transplants, and have a burn unit.

    Hospital prices vary considerably within local markets, even after adjustment for teaching activities and differences in the complexity of cases treated by different hospitals. Two predominant narratives have been offered to explain this disparity: First, high-price hospitals have a special mission and unique characteristics that increase their costs and for which they should be compensated; and second, some hospitals have market power that allows them to negotiate high prices with private health plans and operate under little pressure to contain costs.

    This study offers support for both narratives. We found that high-price hospitals within markets were more likely to be engaged in medical education; offer specialized, expensive services typically associated with tertiary care hospitals; and serve a higher percentage of low-income (and poorly reimbursed) patients. But high price hospitals also clearly enjoyed dominant market positions. Both their large size and their membership in even larger hospital systems made it difficult for health plans to negotiate lower prices with them.

    The authors conclude that the findings highlight “speed bumps” in “attempts to steer patients away from higher-priced providers.” Excluding from networks high-reputation hospitals that offer specialized services could harm enrollees’ access to the care they need and expect. Potentially promising and different approaches includes the use of reference pricing, though that can only work for services that are provided at a wide range of hospitals.

    As James Robinson and Timothy Brown suggest, reference pricing may help payers counter the growing market power of high-clout providers.

    * This point is disputed.


    Comments closed
  • A test survey

    I’m testing a survey plug-in SurveyMonkey, to see if I want to use it for a future survey of readers. Feel free to take this survey, even though it’s just a test. That’d be helpful, actually, so I can see if there are any issues in collecting data.

    Here’s a test survey question: Please rate the letters “A”, “B”, “C” in order of preference. 1 = most preferred letter. 2 = middle preference. 3 = last preference. If it helps, swap in your own question by associating the letters with three other things, like types of pie, brands of coffee, fonts, kind of pet, or whatever.

    You can take this survey below, in the embedded SurveyMonkey thing, or you can go to this link, which may work better, depending on your device.

    You should know that there are some ways in which this survey technology works over which I have no control. For example, with exceptions for some devices (see below), the way you have to rank these is by dragging and dropping them in order, not with the drop down menu next to each choice. If you’re trying to use the drop down menus and they’re auto-filling, drag and drop the choices instead. I’ve been told by email that this may not work on an iPad, but others say it does. Also, on Android devices, I’m told the drop down menus work, but not dragging choices. If you’ve found otherwise, or other devices on which it fails, let me know.

    Comments open for feedback.

    Create your free online surveys with SurveyMonkey , the world’s leading questionnaire tool.


    Comments closed
  • An honest methods section

    By PhD Comics:

    honest methods


    Comments closed
  • Disclosing drug and alcohol related data

    I’ve heard from trustworthy sources that in its Medicare and Medicaid research data files, CMS is no longer including claims for patients with certain alcohol and drug use disorder related diagnoses and procedures. I have not found anything online that documents this more fully. I will update this post should anything come to my attention.

    Here’s an email from ResDAC:

    email ResDAC

    Suffice it to say, lack of access to such data for research is harmful to work on these specific populations, but also any other work, since it imposes a systematic bias on the sample. It messes up risk adjustment too.

    Think about it. Think about the precedent it sets as well.

    For this reason, I reviewed this entry in the code of federal regulations. At the time I accessed that link, the page said the information was current as of December 3, 2014. I’m having trouble assessing if it was published earlier than that date. I’ve excerpted some key passages below. I’ve added bold to some text. With regard to the issue at hand, I think it speaks for itself. (Update: Maybe they don’t speak for themselves as clearly as I thought. You should read this Nicholas Bagley post for an interpretation of the regulations. All posts on this issue are under the CMS-SUD tag.)

    From section 2.1, subsection (a):

    Records of the identity, diagnosis, prognosis, or treatment of any patient which are maintained in connection with the performance of any drug abuse prevention function conducted, regulated, or directly or indirectly assisted by any department or agency of the United States shall, except as provided in subsection (e) of this section, be confidential and be disclosed only for the purposes and under the circumstances expressly authorized under subsection (b) of this section.

    Subsection (b) includes:

    (2) Whether or not the patient, with respect to whom any given record referred to in subsection (a) of this section is maintained, gives his written consent, the content of such record may be disclosed as follows: […]

    (B) To qualified personnel for the purpose of conducting scientific research, management audits, financial audits, or program evaluation, but such personnel may not identify, directly or indirectly, any individual patient in any report of such research, audit, or evaluation, or otherwise disclose patient identities in any manner.

    Further on, it reads, in subsection (e):

    The prohibitions of this section do not apply to any interchange of records—

    (1) within the Armed Forces or within those components of the Veterans’ Administration furnishing health care to veterans

    Section 2.2, as near as I can tell, has identical language to 2.1 but pertaining to “alcoholism or alcohol abuse education, training, treatment, rehabilitation, or research.”

    Section 2.3 includes

    (2) These regulations are not intended to direct the manner in which substantive functions such as research, treatment, and evaluation are carried out. They are intended to insure that an alcohol or drug abuse patient in a federally assisted alcohol or drug abuse program is not made more vulnerable by reason of the availability of his or her patient record than an individual who has an alcohol or drug problem and who does not seek treatment.

    From section 2.52:

    (a) Patient identifying information may be disclosed for the purpose of conducting scientific research if the program director makes a determination that the recipient of the patient identifying information:

    (1) Is qualified to conduct the research;

    (2) Has a research protocol under which the patient identifying information:

    (i) Will be maintained in accordance with the security requirements of §2.16 of these regulations (or more stringent requirements); and

    (ii) Will not be redisclosed except as permitted under paragraph (b) of this section; and

    (3) Has provided a satisfactory written statement that a group of three or more individuals who are independent of the research project has reviewed the protocol and determined that:

    (i) The rights and welfare of patients will be adequately protected; and

    (ii) The risks in disclosing patient identifying information are outweighed by the potential benefits of the research.

    (b) A person conducting research may disclose patient identifying information obtained under paragraph (a) of this section only back to the program from which that information was obtained and may not identify any individual patient in any report of that research or otherwise disclose patient identities.

    Finally, if you’re interested, section 2.16 says, in its entirety:

    (a) Written records which are subject to these regulations must be maintained in a secure room, locked file cabinet, safe or other similar container when not in use; and

    (b) Each program shall adopt in writing procedures which regulate and control access to and use of written records which are subject to these regulations.


    Comments closed