• Blogging: Is it good or bad for journal article readership?

    While at the AcademyHealth Annual Research Meeting last week I had several conversations with editors and board members of various journals, among other attendees, about how blog summaries of academic literature change readership of journal articles. Do blog posts broaden access to people who would not otherwise ever know anything about health policy-relevant research? Or do they allow people who might otherwise read academic papers to skirt by without doing so? (And, if so, is that really a bad thing?)

    My guess: both! Let’s face it, almost nobody who isn’t a researcher or a policy wonk is going to read an article in an academic journal. To the extent that a blog summary reaches beyond the rarefied research and policy communities, it extends the reach of the literature and the ideas it conveys. If one is interested in building a case for the broad relevance of health services and health economics research (among other subject areas), this is unambiguously good.

    It is no doubt true, however, that many in the field use a blog summary as a substitute for other ways of engaging the literature, including reading the papers it references. Yet, I submit that few read an entire paper anyway. What people often do is read the abstract, then maybe the introduction and concluding discussion. Perhaps they add to that a light skim of other sections. Very often a blog post includes more about a paper than is in an abstract, and hits many of the points made in a concluding discussion, as well as some that aren’t made. So, though a blog post may be a substitute, it may not be substituting for any less engagement with a paper’s original and related ideas.

    Finally, I am also confident that for some a blog post is a complement to reading the whole thing. Scanning tables of contents for possible papers of interest is, perhaps, the floor of habitual engagement with the literature. Academics and researchers should probably do at least that. Yet, I know from conversations that even this gets overlooked by many. The torrent of literature is so voluminous these days that even keeping up with tables of contents is not so easy for the busiest researchers and academics. At least a post on one’s favorite blog might bring to one’s attention a paper that one really does want to read.

    What I think may worry some journal editors, board members, and other schoars is that blog posts might be “dumbing down” research to reach broader audiences. (I imagine Twitter further heightens this unease, even if it does expand the potential audience.) That’s certainly a worthwhile concern. It is possible to lose valuable nuance when attempting to simplify and interpret. But broadening access need not mean distorting the message. It all depends on how it’s done. I would hope TIE could be (and is) viewed as part of a “solution” to increasing understanding of the value and content of research. I would most certainly be upset if it was (or is) viewed as “distorting” or “dumbing down” research, or somehow as “the enemy” or a “bad influence.” If anyone in the field feels that way, I encourage him or her to bring that to my attention.

    I emailed about this with Nicholas Bagley, who responded

    When it comes to my work, I’m delighted when someone blogs about it. I figure only a tiny sliver of the population has the time to read the whole thing. The chance to expose more people to my ideas is exciting, even if they get just a simpler version of those ideas. And I’m skeptical that those who are really interested in the topic will decline to do so because they’ve read a summary; probably they wouldn’t have read it anyhow.

    Also, if I had to read everything you and Aaron blogged about, I wouldn’t do much else. Reading summaries gives me a breadth of knowledge that orients me when I engage more deeply in a particular set of problems.

    Responding to an early draft, Bill Gardner wrote me,

    I think that blogs can give you free space to think across disciplines and publish things that do not have a home in specialist journals. They also allow you to publish a more science-based commentary on current events than even an op-ed page will allow.

    (See also Bill’s recent post on research translators.) Comments are open on this post for one week so you can weigh in too. Having said that, I’ll be away and off-line for much of the next week, so please excuse the very long delay in posting your comment.


    Comments closed
  • What we know about hospital networks of exchange plans

    McKinsey did some impressive work collecting hospital network data for 2014 exchange (“marketplace”) plans.

    We have [] enhanced our hospital network database to include all products in all tiers in all 501 rating areas in the U.S. [...] Our database includes all 282 payors filing on the 2014 exchanges and all 4,773 acute care hospitals in the U.S. The payors offered a total of 20,818 on-exchange products across the five metal tiers; these products included 2,366 unique individual exchange networks.

    I believe one has to hit and scrape marketplace and/or plan websites repeatedly to collect such data, so it’s either a large, manual job or one requiring some clever programming. I have little doubt that many enterprising graduate students are building or have built a similar database, but McKinsey’s report is the first product based on such a thing that I’ve seen. It’s an interesting read.

    McKinsey categorized networks as broad (“more than 70 percent of all hospitals in the rating area participating”), narrow (“31 to 70 percent of all hospitals in the rating area participating”), and ultra-narrow (“30 percent or less of all hospitals in the rating area participating”). Here are four, of many, findings:

    1. “Broad networks are available to close to 90 percent of the addressable population” and cost 13%-17% more in premium relative to narrow networks.
    2. “There is no meaningful performance difference between broad and narrowed exchange networks based on Centers for Medicare and Medicaid Services (CMS) hospital metrics such as the composite value-based purchase score as well as its three sub-components (outcome, patient experience, and clinical process scores).”
    3. “26 percent of those who indicated they had enrolled in an ACA plan were unaware of the network type they had selected.”
    4. “Among the new entrants, Medicaid payors and provider-based plans offer the highest percentage of ultra-narrow networks (57 percent and 31 percent, respectively).”

    Finding 1 certainly gives the impression that broad networks are widely available, albeit for a price. However, it’s also true that, relative to incumbents’ 2013 individual market offerings, the proportion of plans with broad networks has fallen, as shown below.


    To the extent one believes CMS hospital metrics to be good measures of quality, finding 2 should be of some comfort to those who only have access to or can afford more narrow network plans.

    Finding 3 surprises me. I would have expected a greater percentage of enrollees to have no idea what kind of network their plan offers. Combining findings 2 and 3, maybe we need not be so concerned about potential opacity of network extent and quality. (Important caveat 1: the analysis in the report only applies to hospitals, not to physician networks. Important caveat 2: the survey results may not be accurately generalized given methodology.)

    Finding 4 is what I would expect of Medicaid- and provider-based plans. Medicaid’s relatively low payment rates aren’t going to appeal to a wide network of hospitals. And, one purpose of provider-based plans is to cater to the providers that offer them, even foreclosing access to those providers by other plans.

    I would have liked to have seen a clearer picture of how network extent varies by metal tiers. The silver tier is of particular interest since premiums of plans in that tier drive premium tax credits and it is only for plans in that tier that cost-sharing subsidies are available. It also, as it turns out, is the most popular plan type.

    The report has many other findings and charts. It’s worth a look.


    Comments closed
  • Relative value health insurance

    In our post on The Upshot last week, Amitabh Chandra and I discussed an idea proposed by Professor Russell Korobkin, relative value health insurance (RVHI, though we didn’t use that term in the piece). In a market for RVHI, plans would be transparently ranked according to the value (degree of cost effectiveness) of services they covered. We wrote

    [A] bronze plan could cover hospitalizations and visits to doctors for emergencies and accidents; genetic diseases; and prescription drugs that keep people out of hospitals. A silver plan could cover what bronze plans do but also include treatments a large majority of physicians find useful. A gold plan could be more inclusive still, adding coverage, for instance, for every cancer therapy shown to improve patient outcomes (no matter the cost) as long as it was delivered at a leading cancer center. Finally, a platinum plan could cover experimental and unproven cancer therapies, including, for example, that proton beam.

    Though Korobkin’s paper is, perhaps, the most thorough consideration of this idea, it’s not the first to propose it. Mark Pauly raised it in his Wussinomics paper (covered on TIE here).

    In particular, one could imagine (as I have suggested before) that insurers offer different plans choosing different cost effectiveness thresholds for new technology, and then consumers could pick the plan with the premium and technology level and growth rates that matched their preferences (Pauly 2005). Not gold, silver, and bronze, but slow-mo versus everything latest. This is Enthoven’s ideal model of managed competition, but it has never really happened. To be sure, there are bargain basement HMOs that will give you a modestly lower premiums than the slap-on-the-wrist PPO but, apart from varying the size of networks, plans have never systematically varied other dimensions of care, like the amount and form of new technology, and competed vigorously on that basis. Instead they waste their time trying to get people to exercise and eat less.

    Pauly (2005) is titled “Competition and New Technology” and says a bit more.

    Plans can thus adopt different policies toward new technologies, and consumers who have a choice among plans can select them based on differences in coverage (broadly defined to include not only reimbursement but also rules, limits, and incentives) of new technology, the implied differences in the growth of premiums, and the value that the consumer places on one relative to the other. As long as consumers face premium differentials that reflect cost, they can in principle choose the optimal plan to limit (or not) the use of new technology. Some plans might permit all new technologies to be used without limit; others might limit them. [...]

    For example, [] a consistent strategy would be to set a benchmark value of dollars per QALY and then adopt all new technologies with costs below that level and none above. Setting a lower threshold would yield a lower rate of growth in spending; plans could therefore vary based onwhat level they chose for their threshold. [...] An alternative would be a bottom-up strategy in which the plan set a target level for spending growth and then used cost-effectiveness analysis to choose the set of new technologies whose cost fit within the limit and which maximized the number of new QALYs delivered. [...]

    Having to face trade-offs between better things is preferable to no trade-offs at all. But dealing in a forthright way with the future path of this effort is surely important, and rejuvenated markets with relevant health plan choices could help a lot.

    Both of Pauly’s papers are worth reading in full.



    Comments closed
  • How to ground

    From geekycrap.tumblr.com:



    Comments closed
  • For- vs. not-for-profit insurers

    The question arose on Twitter how for- and not-for-profit insurers differ. There’s at least one paper on this subject (h/t Janet Weiner). [Update: There's another.*] I noted it once in a reading list, but never posted details. Let’s fix that.

    From Does it Matter if Your Health Insurer is For-Profit? Effects of Ownership on Premiums, Insurance Coverage, and Medical Spending, by Leemore Dafny and Subramaniam Ramanarayanan (NBER, 2012):

    In this study, we use a large, national panel dataset on employer-provided insurance between 1997 and 2009 to study the effect of ownership status on self and fully-insured premiums. We supplement these data with state-level measures of insurance coverage and medical loss ratios. [...]

    A 27-percentage point increase in local FP [for profit] share (one standard deviation) is predicted to raise fully-insured premiums by roughly 7 percent; the effect on self-insured premiums is smaller and cannot consistently be distinguished from zero. Importantly, we do not observe different pre-conversion price trends in markets ultimately experiencing conversions relative to markets whose BCBS affiliates attempted but failed to convert.

    [W]e find heterogeneous effects of conversions in markets with different degrees of BCBS activity. Specifically, we estimate that fully-insured premiums increased roughly 13 percent when converting BCBS plans had shares in excess of the mean pre-conversion BCBS share (20% in our sample), and roughly zero when pre-conversion share fell below the mean. Consistent with oligopolistic pricing behavior, price changes in markets with high preconversion BCBS share were similar for both BCBS and its rivals. It is possible that quality improvements “warranted” the price increases, but we find this explanation somewhat implausible given the similarity in price changes across all insurers. While converting plans underwent major overhauls during which quality improvements could have been implemented, rivals (in general) did not. One would have to believe that rivals made quality improvements of essentially the same market value as BCBS in all markets, i.e. greater improvements where BCBS was relatively more dominant and smaller improvements where BCBS was smaller. Given the challenges associated with generating and marketing changes in quality, as well as the fact that most rivals to BCBS in our sample are national firms, we conjecture that quality improvements likely did not account for all of the observed price increases following conversions. [...]

    [T]he findings have several implications for regulatory and competition policy vis-à-vis insurers. First, it appears that sizeable FP insurers are more likely to exercise market power via price increases than are comparable NFP [not-for-profit] insurers. Second, pricing actions by dominant insurers have a ripple effect on rivals’ prices, further solidifying the evidence pointing towards oligopolistic conduct in many local insurance markets. Third, there is no evidence that NFP and FP insurers charge different prices in the large group market when both are relatively small. These findings suggest that subsidies for de novo NFP insurers (such as those included in the Affordable Care Act) are likeliest to generate value if they facilitate the creation of relatively large players.

    Here are two prior posts on for- vs not-for-profit hospitals and nursing homes.

    * UPDATE:


    Comments closed
  • JAMA Forum: Obamacare’s failure to fulfill its moral ambition

    A health policy goal among some on the left is universal coverage. On the right, the emphasis is often affordable coverage, even if it’s only for some (like those with continuous coverage). The center, in my view, is universal access to affordable coverage. And that just happens to be the best characterization of the main ambition of Obamacare.

    Can we justify this ambition with more than its claim to the political center? I think we can. In a new post on the JAMA Forum I articulate a moral argument for universal access to affordable health insurance, borrowing from the work of Daniels, Saloner, and Gelpi and Saloner and Daniels. The starting point is an assertion that we have a moral obligation to protect opportunity, access to health care being one necessary condition for it.

    Access to health care is enhanced by health insurance. As Daniels, Saloner, and Gelpi argue, universal health insurance is a means to this end. But it’s not the only way. The key is to recognize that equality of access is not equality of receipt. The authors are not suggesting that we have a moral obligation to ensure that everyone receive the same amount of health care, merely that everyone have the same degree of access to it.

    This more modest obligation would be met in a system that does not cover everyone but extends equal opportunity of access to affordable coverage to everyone. That is, equal opportunity to obtain coverage is a necessary condition for equal access to health care, though some may choose not to avail themselves of that care or that coverage. Put another way, if we are morally satisfied with a regime under which people can choose whether to receive care, we ought to be morally satisfied with one under which people can choose whether to obtain coverage for it, so long as there is equal opportunity of access to that coverage and the care it facilitates.

    So much for a moral justification of the law’s ambition, what about its actual implementation? Here it fails; universal access to affordable coverage has not been obtained and is not expected to be under the law. Millions of poor residents in states not expanding Medicaid lack it. And, as Jed Graham of Investor’s Business Daily recently reported, some families covered by exchange plans could face out-of-pocket costs as high as 40%. One can hardly call that affordable.

    Obamacare’s ambition may have a reasonable, moral foundation, but it has not fulfilled it. I wrap up the post with some suggestions on how to better align policy with what morality demands of it. Go read the whole thing.


    Comments closed
  • Maybe it’s the guns

    Lyrics of “If it were up to me,” by Cheryl Wheeler:

    Maybe it’s the movies, maybe it’s the books
    Maybe it’s the bullets, maybe it’s the real crooks
    Maybe it’s the drugs, maybe it’s the parents
    Maybe it’s the colors everybody’s wearin
    Maybe it’s the President, maybe it’s the last one
    Maybe it’s the one before that, what he done
    Maybe it’s the high schools, maybe it’s the teachers
    Maybe it’s the tattooed children in the bleachers
    Maybe it’s the Bible, maybe it’s the lack
    Maybe it’s the music, maybe it’s the crack
    Maybe it’s the hairdos, maybe it’s the TV
    Maybe it’s the cigarettes, maybe it’s the family
    Maybe it’s the fast food, maybe it’s the news
    Maybe it’s divorce, maybe it’s abuse
    Maybe it’s the lawyers, maybe it’s the prisons
    Maybe it’s the Senators, maybe it’s the system
    Maybe it’s the fathers, maybe it’s the sons
    Maybe it’s the sisters, maybe it’s the moms
    Maybe it’s the radio, maybe it’s road rage
    Maybe El Nino, or UV rays
    Maybe it’s the army, maybe it’s the liquor
    Maybe it’s the papers, maybe the militia
    Maybe it’s the athletes, maybe it’s the ads
    Maybe it’s the sports fans, maybe it’s a fad
    Maybe it’s the magazines, maybe it’s the internet
    Maybe it’s the lottery, maybe it’s the immigrants
    Maybe it’s taxes, big business
    Maybe it’s the KKK and the skinheads
    Maybe it’s the communists, maybe it’s the Catholics
    Maybe it’s the hippies, maybe it’s the addicts
    Maybe it’s the art, maybe it’s the sex
    Maybe it’s the homeless, maybe it’s the banks
    Maybe it’s the clearcut, maybe it’s the ozone
    Maybe it’s the chemicals, maybe it’s the car phones
    Maybe it’s the fertilizer, maybe it’s the nose rings

    Maybe it’s the end, but I know one thing.
    If it were up to me, I’d take away the guns.


    Comments closed
  • The food pyramid’s third dimension

    An excerpt from “Lies I Will Tell My Children“:

    food pyramid 3D


    Comments closed
  • In praise of the NNT

    First read this, adapted from a paper by Cook and Sackett:

    Example 1: Relative to a control group, for patients with mild hypertension, the therapy reduced the absolute risk of stroke over the next five years by 0.006. For patients with moderate hypertension, the therapy reduced it by 0.08. [Assume "mild" and "moderate" hypertension are appropriately defined elsewhere.]

    Next, what’s your knee-jerk response to, “How well does this therapy work?” If you’re anything like me, you’re immediately trying to convert the numbers above into something more intuitive. For example, it’s fairly clear that for moderate hypertensive, the therapy would spare 8 out of 100 patients a stroke in the next five years (because 0.08 = 8%). For mild hypertensives, my quick mental calculation would be that the therapy did so for about 1 out of 200 patients (since 0.006 is close to 0.5%).

    Now we’re getting somewhere. I have a decent, intuitive feel for what 8 out of 100 and 1 out of 200 mean as numbers. But we can do a bit better. Instead of reporting the absolute risk reduction, we could report the number needed to treat (or NNT). Try this:

    Example 2: Relative to a control group, the number needed to treat (NNT) to spare one mild hypertensive patient a stroke over the next five years is 167. The NNT for moderate hypertensives is 13.

    The NNTs 167 and 13 are just the (rounded) reciprocals of the absolute risk reduction 0.006 and 0.08 from the first example above. What they’re telling you is that one individual with mild hypertension out of 167 treated will benefit; one out of 13 moderate hypertensives will. Aren’t they easier to interpret? I think so.

    Put yourself in the place of the patient here. Are you the one who will benefit or not? If you’re a mild hypertensive and you’re the one who benefits, then 166 of your fellow patients are the unlucky ones who don’t. That’s a lot of therapy doing nothing! If you’re a moderate hypertensive and you benefit, 12 of your fellow patients don’t. The odds are much better in this case and less therapy is wasted.

    What I haven’t told you is that the numbers above are from studies that show that the relative risk reduction provided by therapy is the same for both groups, 40%. What absolute risk reduction (Example 1) makes clear, and the NNT (Example 2) does even better, is that the absolute benefit of therapy is very different for the two populations, even though the relative benefit is the same. It’s the absolute benefit that matters.

    If you’re surprised by the NNTs in the example above—perhaps thinking they’re not typical—this is an important moment for you. Though single digit NNTs exist, NNTs in the tens to hundreds (or greater) are very typical in medicine. A ton of stuff gets done without any benefit because we have a hard time focusing treatment on precisely those for whom it will definitely help. I bet you think every time you pop a pill or get a procedure you’re benefiting. Guess what? The chances you’re not are very, very high. Are you the lucky one or not?

    I plan to say more about NNTs later in the summer. For now, just let the idea, and their typical size, sink in. Here are some more resources:


    Comments closed
  • The effect of Medicaid on educational attainment

    From The Effect of Child Health Insurance Access on Schooling: Evidence from Public Insurance Expansions, by Sarah Cohodes, Samuel Kleiner, Michael Lovenheim, and Daniel Grossman (NBER, 2014):

    The effect of Medicaid expansions on access to healthcare and on subsequent child health has been studied extensively, (e.g., Currie and Gruber, 1996a, 1996b; Moss and Carver, 1998; Baldwin et al., 1998; Cutler and Gruber, 1996, LoSasso and Buchmueller, 2004; Gruber and Simon, 2008), typically showing that Medicaid expansions increase healthcare access, decrease infant mortality, and improve childhood health. Furthermore, these expansions and Medicaid access more generally have been linked to a lower likelihood of bankruptcy and to less medical debt (Gross and Notowidigdo, 2011; Finkelstein et al., 2012). If Medicaid leads to better health outcomes among children and to more stable finances among low-income households, as suggested by prior research, Medicaid expansions could lead to long-run benefits for affected children.

    But the effect of Medicaid expansion for children on their educational attainment has not been studied, until this paper.

    We find consistent evidence that Medicaid exposure when young increases later educational attainment. A 10 percentage point increase in average Medicaid eligibility between the ages of 0-17 decreases the high school dropout rate by 0.5 of a percentage point, increases college enrollment by between 0.7 of a percentage point and 1.0 percentage point, and increases the four-year college attainment rate (i.e., BA receipt) by 0.9-1.0 percentage point. These estimates translate into declines in high school non-completion of about 5%, increases in college attendance of between 1.0% and 1.5% and increases in BA attainment of about 3.3%-3.7% relative to the sample means.

    Go read Adrianna for more on this study.


    Comments closed