• Working around Halbig

    Suppose the D.C. Circuit’s decision in Halbig becomes the law of the land. If that happens, the states with federally established exchanges will come under enormous pressure to establish their own exchanges. In turn, the federal government would want to make it as easy as possible for those states to convert to state-established exchanges.

    Ideally, HHS would also want to relieve states of the need to develop new exchange infrastructure. Rollout challenges in Oregon and Massachusetts, not to mention Healthcare.gov, suggest that getting a website up and running isn’t such a simple task. What if the refusal states could just enact laws (or sign executive orders) saying they’ve “established” their exchanges, but let Healthcare.gov continue to run them?

    Pointing to the text of the ACA, some critics have said that this wouldn’t work. The ACA provides that an exchange must operate through an “eligible entity.” Among other things, an entity is eligible only if it is incorporated under the laws of “1 or more States.” Because the entity that runs Healthcare.gov is federally chartered, it wouldn’t qualify.

    That’s true, so far as it goes. But the text of the ACA leaves enough room for a workaround. A state could, for example, establish an exchange and appoint a state-incorporated entity to oversee and manage it. That state-incorporated entity could then contract with Healthcare.gov to operate the exchange. On the ground, nothing would change. But tax credits would be available where they weren’t before.

    I don’t see any legal obstacle to that approach. Larry Levitt doesn’t either; as he wrote on Twitter, “If Halbig stands, the administration could try to make it easy for states to set up state exchanges with a healthcare.gov back-end.” Switching would be pretty painless.

    True, not every state would accept the invitation to establish its own exchange, even if doing so were more or less a formality. But lots of states would, especially as voters started to howl about losing their tax credits. If so, even a bad outcome in Halbig might not matter that much in the end.


    Comments closed
  • The government may have lost in D.C., but it just won in the Fourth Circuit

    It’s a busy day. Just hours after the D.C. Circuit invalidated an IRS rule extending tax credits to federally established exchanges, the Fourth Circuit issued an opinion upholding the very same rule.

    The Fourth Circuit’s decision in King v. Burwell basically adopts the theory laid out in Judge Edwards’s dissent in Halbig. In the Fourth Circuit’s view, the relevant ACA language—the language that pins the calculation of tax credits to the cost of a plan purchased on an exchange that was “established by the State under 1311”—is “ambiguous and subject to multiple interpretations.”

    The court started its analysis by agreeing that it’s possible to read the ACA to withdraw tax credits from those purchasing health plans on federally established exchanges. “There can be no question that there is a certain sense to the plaintiffs’ position,” the court acknowledges. But the court refused to get bogged down by a “particular statutory provision in isolation.” Context matters for statutory interpretation.

    And the context here, the court held, cuts against the challengers’ interpretation. The ACA also provides, in §1321, that, when a state fails to establish an exchange, the Secretary “shall . . . establish and operate such Exchange within the State.” In the court’s view, “it makes sense to read § 1321(c)’s directive that HHS establish ‘such Exchange’ to mean that the federal government acts on behalf of the state when it establishes its own Exchange.”

    The court also looked to the broader context of the statute, where several provisions would make no sense under the challengers’ interpretation. The ACA requires federally established exchanges to report to the IRS about tax credits offered on their exchanges. But that would be a senseless requirement if tax credits weren’t available. “It is therefore possible to infer from the reporting requirements that Congress intended the tax credits to be available on both state- and federally-facilitated exchanges.”

    Similarly, the court noted that the ACA allows “qualified individuals” to buy health insurance on exchanges, but defines a “qualified individual” to mean someone who “resides in the State that established the Exchange.” In states with federally established exchanges, accepting the plaintiffs’ argument would “leave the federal Exchanges with no eligible customers, a result Congress could not possibly have intended.”

    At the end of the day, the court said that it could not definitively “discern whether Congress intended one way or another to make the tax credits available on HHS-facilitated exchanges.” As such, the court reasoned, under basic principles of Chevron deference, the IRS’s interpretation of the ambiguous statute was owed deference. That’s especially so, the court reasoned, since “the plaintiffs do not dispute that the premium tax credits are an essential component of the Act’s viability.”

    What does this mean, especially in light of the D.C. Circuit’s decision today? Odds are the Fourth Circuit case won’t be taken en banc. Not only did the panel get it right (at least as I see it), but the five Republican appointees are outnumbered by the ten Democratic appointees (including Gregory, who was originally appointed by Clinton and wrote today’s Fourth Circuit opinion).

    If the D.C. Circuit does vote to rehear Halbig en banc, the likely result would be an opinion upholding the tax credits on a theory that looks like the one the Fourth Circuit adopted. That would then be no split, which might diminish the likelihood of Supreme Court review. If the D.C. Circuit panel’s opinion stands, however, Supreme Court review is almost inevitable. Stay tuned.


    Comments closed
  • 4.683 million unanswered questions in Halbig

    Appeals will continue, but let’s take today’s Halbig decision at face value. How much will this decision cost the working poor? The amount varies with income and other variables, but for a 40 year old individual making $30,000 a year, the tax credit was estimated at $1345 (KFF estimate here). Retroactive tax bills under Halbig will be significant and everyone impacted will have trouble paying for health insurance going forward (about 57% of exchange participants were previously uninsured, according to a KFF survey).

    How many people will be hurt?

    At first glance: anyone receiving tax credits in the 27 states with federally facilitated exchanges (FFEs): (AL, AK, AZ, FL, GA, IN, KS, LA, ME, MS, MO, MT, NE, NJ, NC, ND, OH, OK, PN, SC, SD, TN, TX, UT, VA, WI and WY; KFF list here). But the government reports 36 states as having FFEs, including 9 additional states not included on the list above (ID, NM, AR, DE, IL, IA, MI, NH, and WV; more on this below). Using this broader definition, 4.683 million Americans may now have a surprising tax bill and be at risk of losing health insurance, being told retroactively that they didn’t qualify for tax credits after all. The breakdown of the 4.683 million at-risk enrollees by state (based on this ASPE report and these state-specific excel files):

    FFE statesWhy are the additional nine states included? Seven are partnership states and two (ID/NM) have short term agreements.

    Idaho and New Mexico couldn’t set up their IT in time and signed agreements to allow CMS to start up their exchanges until the states are ready to take over. Will 95,156 of their residents (69,780 and 25,376, respectively) have to refund their tax credits until the switch occurs?

    The data from ASPE includes as FFE the seven states with “partnership” exchanges (AR, DE, IL, IA, MI, NH and WV). Do these arrangements count as “an Exchange established by the State?” If so, 527,000 people don’t lose their tax credits today. But if this model works, why can’t any state negatively affected by today’s decision simply sign a quick “partnership” agreement with CMS? If this works prospectively (as was suggested at oral arguments before the 4th Circuit in a related case), what about the tax credits from 2014? Will Halbig just punish about 4.683 million working poor for 2014 – but not 2015 and beyond – if quick agreements are signed by CMS and the states?

    But some state-based exchange states are in flux, which could increase the totals above. Oregon is in the process of switching to a FFE because of an epic website failure. Does that mean that 54,663 Oregon residents lose their tax credits once the switch occurs in November? Maryland is also considering the federal exchange after start up troubles. In Maryland, the number at risk is not separately reported, but could be estimated at 58,271 (86% of 67,757; 86% is the overall percentage of FFE enrollees qualifying for financial assistance). No one has told the working poor in Oregon and Maryland that over 110,000 people could lose these tax credits.

    And what about states that might outsource their exchange to another state? (MA, MD, MN and NV are considering this). Does an exchange operated by another state qualify as “an Exchange established by the State” under IRC sec. 36B(b)(2)(A) (emphasis added)? Seem likely, but if not, up to another 170,000 are at risk.

    Some special situations deserve mention. In Utah and Mississippi, the state runs the SHOP exchange and the federal government runs the individual exchange. Does that mean that if you are insured through a small business in Biloxi or Provo you keep your tax credits, but if you purchased as an individual you owe a tax bill?

    In addition, consider the seven “plan management” states? (KS, ME, MT, NE, OH, SD, VA). Some functions in these FFEs are performed by the states by agreement with CMS. If these ostensible FFE states are actually “established by the State,” then 442,000 people keep their tax credits. (Kansas claimed in amicus briefs that their citizens will lose tax credits, which is inconsistent with this argument).

    ASPE also collected racial data:



    Comments closed
  • A stinging defeat for the government

    In a major setback for the Affordable Care Act, the D.C. Circuit just released a fractured opinion invalidating the IRS’s rule extending tax credits to federally facilitated exchanges.

    The case, Halbig v. Sebelius, centers on the portion of the ACA governing the calculation of tax credits. The statute specifies that tax credits are available to most people who purchase a health plan “through an Exchange established by the State under 1311.” (See my earlier posts for a more detailed recap.) About two-thirds of the states, however, declined to establish exchanges. In those states, the federal government stepped in and established the exchanges on the states’ behalf.

    In today’s opinion, the D.C. Circuit held that a federally facilitated exchange isn’t “established by the State under 1311.” As a result, the IRS can’t offer tax credits to those who purchase plans on such exchanges. Since the average estimated tax credit in 2014 is $4,700, the ruling threatens to deprive tens of thousands of people in Texas, Florida, Ohio, Michigan, and many other states of the means to buy health insurance.

    In his opinion for the Court, Judge Griffith starts with the text of the statute. He first acknowledges that a federal exchange is a 1311 exchange, even if it’s established by the Secretary of HHS under 1321 of the ACA. After all, section 1321 instructs the Secretary to establish “such exchange” if a state fails to do so. In context, “such exchange” clearly refers to a 1311 exchange.

    But that’s not enough. As Griffith sees it, “[t]he problem confronting the IRS Rule is that subsidies also turn on … who established them.” The statutory text requires the exchanges—even those established under 1311—to be “established by the State.” Because federal exchanges aren’t established by a state, but by the federal government, individuals who purchase a plan on federally established exchanges are ineligible for tax credits.

    Griffith then turns to the larger statutory context, and to the government’s claim that a cramped construction of the statute “would render several other provisions of the ACA absurd.” What about the ACA requirement that federally established exchanges report on who receives tax credits? Wouldn’t that be superfluous if no one received any such credits? “Not so,” says Griffith. “Even if credits are unavailable on federal Exchanges, reporting by [federally established] Exchanges still serves the purpose of enforcing the individual mandate.”

    What about the ACA provision stating that “qualified individuals” can buy plans on an exchange? Since a “qualified individual” is defined in the statute to mean someone who “resides in the States that established the Exchange,” Griffith acknowledges that giving this provision its plain meaning would mean that “the 36 states with federal Exchanges … have no qualified individuals.” Even so, he says, “[t]he government … tilts at windmills.” In Griffith’s view, “[t]he obvious flaw in this interpretation is that the word ‘only’ does not appear in the provision.” People in states with federally facilitated exchanges should be allowed on those exchanges, even if the statute might at first glance appear to preclude them from doing so.

    Finally, Griffith addresses the legislative history of the ACA and concludes that it “sheds little light on the precise question on the availability of subsidies on federal Exchanges.” In Griffith’s view, that silence about whether Congress intended the odd result of depriving individuals on federal exchanges of subsidies is not enough. “[T]here must be evidence that Congress meant something other than what it literally said.”

    Griffith concludes his opinion with the following remarkable statement:

    We reach this conclusion, frankly, with reluctance. At least until states that wish to can set up Exchanges, our ruling will likely have significant consequences both for the millions of individuals receiving tax credits through federal Exchanges and for health insurance markets more broadly. But, high as those stakes are, the principle of legislative supremacy that guides us is higher still. Within constitutional limits, Congress is supreme in matters of policy, and the consequence of that supremacy is that our duty when interpreting a statute is to ascertain the meaning of the words of the statute duly enacted through the formal legislative process. This limited role serves democratic interests by ensuring that policy is made by elected, politically accountable representatives, not by appointed, life-tenured judges.

    In a lengthy and passionate dissent, Judge Edwards notes his disagreement at every turn with the majority:

    The majority opinion ignores the obvious ambiguity in the statute and claims to rest on plain meaning where there is none to be found. In so doing, the majority misapplies the applicable standard of review, refuses to give deference to the IRS’s and HHS’s permissible constructions of the ACA, and issues a judgment that portends disastrous consequences. I therefore dissent.

    What happens now? Instead of taking the case right to the Supreme Court, the government will probably ask the whole D.C. Circuit to review it. (The government has until September 5 to file its petition.) The court is very likely to review the case en banc: it’s undeniably of “exceptional importance” and the decision is, in my view, quite wrong. It also won’t hurt that, after filibuster reform, the court’s seven Democratic appointees outnumber its four Republican appointees.

    In all likelihood, the case would be heard en banc in the late fall or winter. If the government loses again—which is unlikely, in my view—the Supreme Court would almost certainly take the case. If the government wins, it’s more difficult to hazard a prediction. Much will depend on how the other pending cases presenting the same question develop, especially King v. Sebelius, which was recently argued before the Fourth Circuit.

    This is by no means is this the final word on the exchange litigation. But that’s not to minimize the significance of the court’s decision today. It lends plausibility to the challengers’ arguments, gives momentum to the litigation—and increases the odds that millions of Americans won’t be able to afford health insurance.


    Comments closed
  • I bet you never heard of “Rule Out Sepsis”. It’s the bane of pediatric interns everywhere.

    So for most kids, you get a fever, and maybe it lets you stay home from school. But in a small infant, it’s panic inducing. That’s because there’s a small, but greater than zero chance that a baby has a serious bacterial infection that could kill it without treatment.

    When I was an intern, any infant under 2 months of age got rushed to the hospital. Then we drew blood for a CBC and blood culture. We stuck a catheter into their bladder to get urine for a urinalysis and urine culture. Then we stuck a needle into their back to obtain spinal fluid to check for cells and do a CSF culture.

    Every infant. Every fever.

    And then we put them in the hospital for 48 hours or more until every single culture came back negative. You have to wait, because you don’t get a “negative result”. You just get nothing growing. That’s a negative. But how long do you wait?

    When it’s called a “negative”, we send you home. I can remember no children who were otherwise well appearing who wound up having a positive culture. But it can happen. So we play it safe.

    As you can imagine, this costs money. It disrupts families. It leaves infants exposed to other illnesses in the hospital. It just plain sucks. By the time I was a fellow, we were trying to develop new guidelines by which we could send some older infants home on antibiotics and follow them from there. But it still sucks.

    So I was happy to see this paper, “Blood Culture Time to Positivity in Febrile Infants With Bacteremia“:

    Importance: Blood cultures are often obtained as part of the evaluation of infants with fever and these infants are typically observed until their cultures are determined to have no growth. However, the time to positivity of blood culture results in this population is not known.

    Objective: To determine the time to positivity of blood culture results in febrile infants admitted to a general inpatient unit.

    Design, Setting, and Participants: Multicenter, retrospective, cross-sectional evaluation of blood culture time to positivity. Data were collected by community and academic hospital systems associated with the Pediatric Research in Inpatient Settings Network. The study included febrile infants 90 days of age or younger with bacteremia and without surgical histories outside of an intensive care unit.

    Exposure:s Blood culture growing pathogenic bacteria.

    Main Outcomes and Measures: Time to positivity and proportion of positive blood culture results that become positive more than 24 hours after placement in the analyzer.

    Results: A total of 392 pathogenic blood cultures were included from 17 hospital systems across the United States. The mean (SD) time to positivity was 15.41 (8.30) hours. By 24 hours, 91% (95% CI, 88-93) had turned positive. By 36 and 48 hours, 96% (95% CI, 95-98) and 99% (95% CI, 97-100) had become positive, respectively.

    Conclusions and Relevance: Most pathogens in febrile, bacteremic infants 90 days of age or younger hospitalized on a general inpatient unit will be identified within 24 hours of collection. These data suggest that inpatient observation of febrile infants for more than 24 hours may be unnecessary in most infants.

    They looked at data on 17, yes seventeen, different hospital systems for between 2 and 6 years. And in all that time, they found only 392 blood cultures from non-ICU or inpatient settings that were positive with pathogens. So this ain’t common. More important, when they were positive, they were found to be so within 15 hours on average. More than 90% of them were positive in a day.

    In other words, we can probably make a judgement call earlier. If nothing grows by 24 hours, and a baby looks well, you can probably go home with strict instructions. This may not sound like a big deal to you, but it would be a huge deal for the doctors caring for these babies, and – much more importantly – the families who just want to get home and be with their newborns.

    Comments closed
  • Methods: Propensity scores


    Forthcoming in Health Services Research (and available now via Early View), Melissa Garrido and colleagues explain propensity scores. I’ve added a bit of emphasis on a key point.

    Propensity score analysis is a useful tool to account for imbalance in covariates between treated and comparison groups. A propensity score is a single score that represents the probability of receiving a treatment, conditional on a set of observed covariates. [...]

    Propensity scores are useful when estimating a treatment’s effect on an outcome using observational data and when selection bias due to nonrandom treatment assignment is likely. The classic experimental design for estimating treatment effects is a randomized controlled trial (RCT), where random assignment to treatment balances individuals’ observed and unobserved characteristics across treatment and control groups. Because only one treatment state can be observed at a time for each individual, control individuals that are similar to treated individuals in everything but treatment receipt are used as proxies for the counterfactual. In observational data, however, treatment assignment is not random. This leads to selection bias, where measured and unmeasured characteristics of individuals are associated with likelihood of receiving treatment and with the outcome. Propensity scores provide a way to balance measured covariates across treatment and comparison groups and better approximate the counterfactual for treated individuals.

    Propensity scores can be thought of as an advanced matching technique. For instance, if one were concerned that age might affect both treatment selection and outcome, one strategy would be to compare individuals of similar age in both treatment and comparison groups. As variables are added to the matching process, however, it becomes more and more difficult to find exact matches for individuals (i.e., it is unlikely to find individuals in both the treatment and comparison groups with identical gender, age, race, comorbidity level, and insurance status). Propensity scores solve this dimensionality problem by compressing the relevant factors into a single score. Individuals with similar propensity scores are then compared across treatment and comparison groups.

    Propensity scores are a useful and common technique in analysis of observational data. They are, unfortunately, sometimes misunderstood as a way to address more types of confounding than they are capable. In particular, they can only address confounding from observable factors (“measured” ones, in the above quote). If there’s an unobservable difference between treatment and control groups that affects the outcome (e.g., genetic variation about which researchers have no data), propensity scores cannot help.

    It is important to keep in mind that propensity scores cannot adjust for unobserved differences between groups.

    Only an RCT or, with assumptions, natural experiments and instrumental variables approaches can address confounding due to unobservable factors. I will return to this issue.

    I’m deliberately not covering implementation issues and approaches in these methods posts, just intuition, appropriate use, and issues of interpretation. If you want more information on propensity scores, read the paper from which I quoted or search the technical literature. Comments open for one week for feedback on propensity scores or pointers to other good methods papers.


  • The US is not making progress on public health

    Aaron recently called our attention to the Healthy People 2020 report on progress in public health. The high-level goals are fundamental contributors to human well-being:

    • Attain high-quality, longer lives free of preventable disease, disability, injury, and premature death;
    • Achieve health equity, eliminate disparities, and improve the health of all groups;
    • Create social and physical environments that promote good health for all; and
    • Promote quality of life, healthy development, and healthy behaviors across all life stages.

    To make progress on these high-level goals, public health experts chose 26 specific and measurable goals, for example, to reduce the number of persons with diagnosed diabetes whose A1c value is greater than 9%.

    The point of setting goals is to track whether we are getting anywhere. Are we?

    Dr. Howard Koh is the Assistant Secretary for Health for the U.S. Department of Health and Human Services. In JAMA, he reports on recent progress (or lack thereof) toward the Healthy People 2020 goals.

    The data demonstrate areas of both improvement and continued need. On the positive side, 14 of the 26 Leading Health Indicators (54%) have documented improvement, and 4 have met or exceeded their Healthy People 2020 targets… The new data also document no improvement for 11 of the 26 indicators (42%) and, of those, 3 show worsening health outcomes.

    We aren’t making meaningful progress on public health. For example, there was little improvement in the percent of persons with high levels of A1C. There was progress in maternal and child health: Infant deaths per 1,000 live births decreased from 6.7 in 2006 to 6.1 in 2010 and total preterm live births declined from 12.7% in 2007 to 11.5% in 2012.

    But we lost ground on both of our national mental health goals: The age-adjusted suicide rate increased from 11.3/100,000 in 2007 to 12.1 in 2010 and the percent of adolescents with major depressive episodes increased from 8.3% in 2008 to 9.1% in 2011.

    Fourteen successes on 25 goals is not statistically better than a 50% success rate (change could not be measured for one of the goals).

    Why aren’t we doing better? One reason is that we don’t pay enough attention to public health data. Every month, the economics wonks wait anxiously for reports on employment, economic growth, and so on. Twitter buzzes about relatively small improvements or declines. Those indicators matter, a lot. But are they more important than longer lives free of preventable disease, disability, injury, and premature death?


    Comments closed
  • AcademyHealth: How does malpractice reform affect physician migration?

    We often hear about malpractice reform pitched in terms of spending reductions. But that ignores its other potential effects. Does it affect physician migration? In a good way or a bad way? There’s a new study on that topic, and I discuss it over at the AcademyHealth blog.

    Go read!


    Comments closed
  • Methods: Intention-to-treat


    In JAMA, Michelle Detry and Roger Lewis explain the “intention-to-treat” (ITT) principle:

    [I]n a trial in which patients are randomized to receive either treatment A or treatment B, a patient may be randomized to receive treatment A but erroneously receive treatment B, or never receive any treatment, or not adhere to treatment A. In all of these situations, the patient would be included in group A when comparing treatment outcomes using an ITT analysis. Eliminating study participants who were randomized but not treated or moving participants between treatment groups according to the treatment they received would violate the ITT principle.

    Why do this?

    The effectiveness of a therapy is not simply determined by its pure biological effect but is also influenced by the physician’s ability to administer, or the patient’s ability to adhere to, the intended treatment. The true effect of selecting a treatment is a combination of biological effects, variations in compliance or adherence, and other patient characteristics that influence efficacy. Only by retaining all patients intended to receive a given treatment in their original treatment group can researchers and clinicians obtain an unbiased estimate of the effect of selecting one treatment over another.

    Treatment adherence often depends on many patient and clinician factors that may not be anticipated or are impossible to measure and that influence response to treatment.

    Why not do this?

    [1] Noninferiority trials, which are designed to demonstrate that an experimental treatment is no worse than an established one, require special considerations. [...] The intervention in group A may incorrectly appear noninferior to the intervention in group B, simply as a result of nonadherence rather than because of similar biological efficacy. [...]

    [2] Although the ITT principle is important for estimating the efficacy of treatments, it should not be applied in the same way in assessing the safety (eg, medication adverse effects) of interventions. [...]

    [3] [I]t would be unfortunate to falsely conclude, based on the ITT analysis of a phase 2 clinical trial, that a novel pharmaceutical agent is not effective when,in fact, the lack of efficacy stems from too high a dose and patients’ inability to be adherent because of intolerable adverse effects. In that case, a lower dose may yield clinically important efficacy and a tolerable adverse effect profile.

    In these cases, one may be more interested in an estimate of the effect of treatment-on-the-treated (TOT), or a per-protocol analysis.

    If you’re aware of good papers that explain the use and interpretation of common research methods, let me know in the comments, which are open for one week after this post’s time stamp, or by email or Twitter.


  • Healthcare Triage: Sunscreen

    Another great episode for summer:

    When I was a kid, I remember people talking about putting on suntan lotion to help them absorb the sun’s rays. Those days are over. Too much sun can be terrible for you. Besides the fact that it significantly increases the risk of skin cancer, the sun will age your skin and make you look older, too. Who wants that? Today, we use sunscreen to protect us from the sun. But most of us are doing it wrong. How so? Watch and learn.

    For those of you who came here for references or more information, try these:


    Comments closed