Healthcare Policy

SMU Commencement Weekend, Part 1 - Funding Scientific Research

Francis-S-CollinsNIH has some groundbreaking ideas to transform science in academia. Read more to learn why. First, the background: tomorrow is Commencement at SMU and NIH Director Francis S. Collins will deliver the Commencement address as well as receive a Honorary Doctor of Science. Three other outstanding individuals will also receive honorary degrees - astrophysicist Francis Halzen of UW-Madison, arts philanthropist Nancy Nasher and New Testament scholar E.P. Sanders (read more about them here). Yesterday and today at SMU saw several remarkable events featuring those awardees.

For this post, I want to focus on the last one, which was a panel discussion involving NIH Director Francis S. Collins, SMU Provost Steven Currall, SMU Professor Pia Vogel and UT Southwestern Medical Center President Daniel Podolsky. Dr. Collins made a presentation about science discoveries that he hopes will happen within 10 years, and then took part in the panel discussion, and finally took questions for the audience. The most interesting one was the last one, about the funding of academic research in science.

Dr. Collins cited a 2014 New York Times op-ed by Andy Harris, Young, Brilliant and Underfunded, that pointed out that most of the Nobel Prize winners and other notable scientists came up with their breakthrough ideas between the age of 35 and 39, "yet the median age of first-time recipients of R01 grants, the most common and sought-after form of N.I.H. funding is 42 while the median age of all recipients is 52. More people over 65 are funded with research grants than those under age 35." 

This was after a comment by Dr. Collins about the N.I.H. having gotten better at funding early-career researchers by putting them in their own, separate pool if they have never received N.I.H. funding before, but not better at funding mid-career researchers, who report getting "squeezed". The aging of science's principal academic investigators is also problematic for the long-term vigor of the field. 

This helps put his next remark in context: according to Dr Collins, recent data suggests the productivity of a N.I.H. principal investigator begins to drop after the third concurrent grant, and that if rules were put in place that reassign funding dollars from those 4th or more concurrent grants, N.I.H. would be able to make 900 extra grants to early- and mid-career researchers (I suppose those grant amounts would be smaller than the grants of the "big shots" but he did not discuss that). He made it clear that they would be for grant proposals that fell just short of funding under the present rules - grant proposals that deserved funding but could not be funded due to insufficient funds.

The other transformative concept Dr. Collins talked about was of "early-independence awards" to help young PhDs skip post-docs and get them "unleashed" earlier, so that they can be creative and make independent groundbreaking discoveries earlier. Not everybody needs a post-doc. What I found most staggering about it is that the fields of engineering and management, where you did not use to need a post-doc to get a faculty position, have slowly become so risk-averse (reluctant to hire just-graduated PhDs in case they don't manage to become independent) that they have aligned themselves more and more on science and now it is quite usual to do a post-doc before obtaining a faculty position, and now that science has led us into a three-stage academic model of PhD/post-doc/faculty position, it is moving away from that. So maybe engineering and management will return to their old ways too.

I think it is particularly welcome for science to develop ways to bypass postdocs because faculty members in science often have lower pay than their colleagues in management or engineering, in addition to often having longer time-to-completion in the PhD program, so they can't easily make up for their lost wages once they are on the faculty. Some even take 1 or 2 years before they apply to PhD programs to work as lab techs (when they plan to go into experimental fields). Others take 2 postdocs before they go on to faculty positions.

All in all we are looking at talented scientists who, in the current model, first become independent Assistant Professors around the age of 35, which is (1) really late to start giving some stability to scientists, and can help explain why many prefer careers outside academia, (2) the low end of the 35-39 age window for the breakthrough discoveries by Nobel Prize winners or equivalent discussed above. You can't really expect brand new Assistant Professors to make that level of discoveries within months of their first faculty position. And of course not everyone will end up a Nobel Prize winner or equivalent, but that's not a reason to needlessly discourage people from staying in academia. There is a case to be made that the thinkers most capable of transformative innovation aren't necessarily always the ones well-established research behemoths but may well be, sometimes, researchers at less well-established institutions. Their ideas deserve to be given a chance too.

My next post will summarize the rest of the panel discussion.

Costs of Health Plans in Public Health Exchanges in PA

Last month I was talking to an employee at my eye doctor (the woman who adjusts customers' glasses) and she mentioned how worried she was by the increase in premiums and deductibles on the Pennsylvania Health Exchange, where she gets coverage and where the plan she has had for the past two years is being discontinued.

When she told me with tears in her eyes that the cheapest plan had a premium of about $300 (which she said was two or three times her current premium, if I understood her correctly) and a deductible of $6,000 I was understandably shocked. She had no idea how she would be able to afford a new plan and obviously doesn't have $6,000 lying around if she gets sick. She is also a very inspirational woman, having broken the gender barrier on her local first-responders' team several decades ago and being so committed to her community that she continues to volunteer for them in a supporting capacity, in addition to other things that make her a really amazing woman. You would think politicians from both sides of the aisle would make it affordable for people like her to purchase health insurance.

It is one thing to know at an intellectual level that the middle class is being decimated in this country but quite another to be staring at an exemplar who doesn't know how she can possibly afford the premiums for the health insurance she has to get. Further, it motivates people who struggle financially to further postpone receive expensive treatment. This cannot have a good outcome.

This got me curious to see what her options were. I went on the site, picked Pennsylvania as the state and made guesses on her age and zip code. I knew that she lives alone and has no dependent (her children are grown). I picked a salary range that wouldn't give her a tax rebate since she mentioned she made "too much money" for the rebate. And sure enough, the cheapest plan, a Bronze HMO from Keystone called "Healthy Benefits Value HMO 6300.50", has a monthly premium of $296 and a deductible of $6,300, which is very steep for the area we live in, especially for people in such a situation that they must get health insurance through the exchange. I suppose people are supposed to be able to afford that thanks to Health Savings Accounts but to many it seems that they have to pay a lot of money every month and then when they actually have health problems and have to use their insurance they still look at a hefty deductible before the insurance kicks in.

The plans on the website are ranked by increasing order of premium but this year the site also provides cost estimates for three scenarios (low, medium and high healthcare costs). 

  • Low: 4 doctor visits, 1 lab or diagnostic tests, 7 prescription drugs, $100 in other medical expenses,
  • Medium: 8 doctor visits, 3 lab or diagnostic tests, 15 prescription drugs, $400 in other medical expenses,
  • High: 20 doctor visits, 11 lab or diagnostic tests, 45 prescription drugs, 2 days in the hospital, $17,700 in other medical expenses,

What surprised me the most is that the Keystone plan with the cheapest deductible is also the best deal if you anticipate low or medium health care costs. I say it surprised me because it is a bronze-level plan and it is common wisdom that you should buy a silver-level plan instead. But if you don't anticipate high healthcare costs, maybe you should rethink. (Of course we can argue about what is a medium cost and what is a high cost, but the website lets you change the scenario definitions of each. My hunch is that high costs involve spending the night in the hospital.)

Also, the 11 plans with the cheapest premiums as well as among the cheapest costs for low, medium and high healthcare costs are all HMOs. They held their ground even for high healthcare costs, although they tended to be not as cheap for the customer as a plan selected specifically with high healthcare costs in mind. The plan that seemed to offer the best all-around value based on this admittedly cursory look was "Healthy Benefits Value HMO 3500.0", a Silver HMO plan from Keystone. This was also the "optimal" plan (cost-wise) to select if the prospective enrollee knew from the beginning that his healthcare costs next year would be high. Its premium, however, is $62 more expensive per month than the cheapest premium, representing a 21% increase, which might dissuade some.

Ultimately, the five cheapest healthcare plans for patients who know their healthcare costs would be high were all Keystone plans. This of course raises questions of what makes Keystone capable of offering so cheap rates (scale? efficiency? contracts with providers?), and when my then PhD student Tengjiao Xiao (now Senior Consultant at Ernst & Young) and myself evaluated health plan efficiency in a previous round of open enrollment, we did have to consider the fact that the cheapest plans may have been cheap to draw in enrollees, including healthy ones, rather than out of a particularly cheap estimate of treating them. Once patients were enrolled, then it has long been thought that they would stick with their health plan even if premiums increased. Yet, this applied to a different segment of the enrollee population, who receives benefits through their employers. People who find coverage through the public health exchanges seem much more willing to shop for deals, in part because many of them are in precarious financial situations. 

You can find the data in this spreadsheet I put together by keying in all the numbers from the result pages. The cells are color-coded per columns with green representing cheapest and red most expensive.

Further reading: some interesting thoughts on trends on the PA health insurance marketplace are provided here.

Pricing of Gilead Sciences drug Sovaldi

Last week the United States Senate Committee on Finance published the results of an 18-month investigation into Gilead Sciences's Hepatitis C drug Sovaldi and its second-wave successor Harvoni. It found that "[the] pricing and marketing strategy was designed to maximize revenue with little concern for access or affordability." Sovaldi cost $84,000 for a single course of treatment and Harvoni $94,500.

This is significant because "Medicare spent more on Gilead Hepatitis C Drugs in the first half of 2015 than in all of 2014". In addition, Medicaid programs spent more than $1 billion to treat less than 2.4 percent of enrolled patients with Hepatitis C with Sovaldi (because patients first try other drugs and Sovaldi only after the other ones fail).

Highlights include:

  • "In 2014 alone, Medicare and Medicaid combined to spend more than $5 billion on Sovaldi and Harvoni before rebates. That total is projected to climb in 2015."
  • "Gilead’s recent financial statements show U.S. sales of Sovaldi and Harvoni, including through public programs and private payers, totaled $20.6 billion after rebates in the 21 months following Sovaldi’s introduction."
  • Senate Finance Committee Ranking Member Ron Wyden, D-Ore., is quoted as saying: “Gilead pursued a calculated scheme for pricing and marketing its Hepatitis C drug based on one primary goal, maximizing revenue, regardless of the human consequences. There was no concrete evidence in emails, meeting minutes or presentations that basic financial matters such as R&D costs or the multi-billion dollar acquisition of Pharmasset, the drug’s first developer, factored into how Gilead set the price. Gilead knew these prices would put treatment out of the reach of millions and cause extraordinary problems for Medicare and Medicaid, but still the company went ahead.
  • "Gilead set a high price for Sovaldi with an eye toward ensuring a future high price for Harvoni." In addition, "by elevating the price for the new standard of care set by Sovaldi, Gilead intended to raise the price floor for all future Hepatitis C treatments, including its follow-on drugs and those of its competitors."
  • "Gilead underestimated the degree of access restrictions that it expected would result from its pricing decision."
  • While prices decreased after competitors entered the market, concerns remain. "Even as competition lowered prices for therapies, this report documents that concerns remain, particularly in the public payer community, about high costs for treating millions of people in the U.S. infected with Hepatitis C, as well as the budgetary effects of a future single source innovator that might not face competition as quickly."
  • The executive summary further mentions that: "while Sovaldi has significantly improved cure rates for the most common variety of Hepatitis C, genotype 1, for other genotypes, rates were lower and required much longer treatment, at a wholesale price as high as $168,000. Gilead did not take meaningful steps to price or discount its drugs to help those individuals." 

You can read the executive summary of the report here and the full report there.

The report also mentions that "Gilead acquired its sofosbuvir-based drugs through the multi-billion dollar acquisition of Pharmasset, Inc. in 2012, and spent hundreds of millions of dollars more completing clinical trials and FDA approvals." Specifically, it bet $11 billion in a "huge and risky bet" (Reuters). So we are also seeing a company that attempts to reward itself for the real financial risk it took back in 2011.

But the Reuters article also states: "Analysts questioned whether the deal price -- equal to more than one-third of Gilead's market value -- was too steep. A Wall Street survey... found that 82 percent thought Gilead paid too much." So perhaps the high price public payers have to pay now follows from the steep acquisition price Gilead agreed to. In fact the founder of Pharmasset stated at the time of the deal that "They could have had the company for $300 million or less in 2004." Maybe public payers shouldn't have to pay for Gilead's mistakes.

While pharmaceutical companies like to justify high drug prices as a way to recoup the R&D costs of drugs that failed and overall as the price people have to pay to maintain a healthy innovation ecosystem, it is sometimes difficult to determine where the price for innovation ends and the price for greed begins.

Highlights from #nhpc15

Here are a few highlights, in no particular order, from the National Health Policy Conference that was held in DC last week.

Elizabeth Bradley, Professor of Public Health at Yale, who co-authored The American Health Care Paradox with and Lauren Taylor of the Harvard Divinity School, talked about population-health approaches. Noteworthy stats: the US, out of 34 OECD countries, is 25th in maternal mortality, 26th in life expectancy and 28th in low birth weight. According to a 2002 Health Affairs paper by McGinnis et al, health is determined at 60% by social, environmental and behavioral factors, while health care makes up only 20% and genetics the remaining 20%.

Bradley investigates spending in (non-health-related) social services as a way to improve health. Social services include employment programs, supportive housing and rent subsidies, nutritional support and family assistance, among others. The US has one of the lowest social-to-health-spending ratios among developed countries; in the graph she showed, the OECD average is slightly over 1.50 while the US ratio is about 0.6. She also points out that, for $1 spent on health care, the US spends $0.90 and the OECD $2 on social services. (I guess the ratio shown on the graph is calculated in some different way, but you get the overall theme.) A possible reason for this discrepancy between the US and the rest of the developed world is that health care grew into a marketable commodity for purchase while social services were viewed "for the poor", as an act of charity.

She suggests incentivizing collaboration on health and gives the examples of:

  1. C-TRAIN in Portland, OR, serving families facing homelessness, poverty and addition,  
  2. the 10th Decile project, targeting the 10% of homeless people with the highest healthcare costs, who were provided permanent housing with strong medical and mental health support. This intervention led to a decrease in healthcare costs by 72% from almost $59k to about $17k.  
  3. the WIN for Asthma project, where bilingual community workers provide asthma education and referrals for housing, immigration and mental health services. This led to a 50% decrease in emergency visits and hospitalizations as well as a 30% decline in school absenteeism.

 Michael Vita, of the Federal Trust Commission, discussed healthcare mergers. (32% of the FTC's enforcement actions over 2009-2013 were in the healthcare sector. The moderator pointed out that onsolidation is a real concern since most urban areas are dominated by 1-3 very large systems, there has been over 350 mergers since 2010 and we have seen a 32% increase in number of doctors employed by hospitals over the last decade, with 20% of physicians now employed by hospitals.) In 2002, the Hospital Retrospective Task Force studied four consummated mergers and found that the Evanston-Highland Park merger had led to a price increase 11 to 18 percentage points above the price of the control group. It successfully challenged the merger, which set the stage for challenge to proposed mergers. While most transactions are approved (the FTC has only challenged less than 1% of the hospital mergers between 2002 and 2012), it has at times challenged mergers of hospitals and mergers of primary care physician practices to avoid market concentration. Vita also discussed FTC rules regarding efficiencies claimed for the upcoming merger and ended with a few words on antitrust and the ACA. ("We see no conflict between goals of ACA and goals of antitrust.") 

Eve Kerr of the VA Center for Clinical Management Research and the University of Michigan Medical School talked about patient-centered performance management, and in particular on goal definition and performance management based on patients' preferences, leading to different care pathways. For instance in the case of diabetes, it is standard practice to monitor hemoglobin A1c levels through the HbA1c test. The goal would be to keep the level below 7%; however, you can't label care as good or bad depending on whether it falls below or exceeds the 7% threshold.

For instance, if the A1c level is between 7% and 7.5%, there will probably be low benefits to expanding high efforts in trying to bring it below 7%. For high levels of A1c, say, above 9%, it is clear that care would yield very high benefits. But patients with a blood level around 8% find themselves more in a gray area in the continuum of care, where there is not a "one-size-fits-all" best care pathway and the best treatment plan would be preference-sensitive. If the patient's goal is indeed to bring the A1c level below 7.5%, he would be willing to try a medication (and thus be more likely to adhere to it); however, if the patient's goal is to keep its A1c level at 8% or less and he is currently right at 8%, he would achieve his goal with a lifestyle modification. While the amount of choice seemed a bit overwhelming and risks creating the need for significant amounts of documentation (how do you code patient preferences in medical records?), in fact this is simply about having multiple well-defined care pathways and letting patients choose the one they feel most comfortable with given their goals, making them more likely to successfully complete treatment.

Michael Chernew of Harvard made brief remarks on payment reform. He discussed traditional fee-for-service, Patient Centered Medical Home type fixed fees, bundled payment and global payment and pointed out a Brookings white paper he recently co-authored, on Medicare Physician Payment Reform: Securing the Connection Between Value and Payment

The NHPC Monday Afternoon Keynote offered many important insights. Jeffrey Kang, Senior VP for Health and Wellness Services and Solutions at Walgreens, talked about retail health, Walgreens' partnership with Theranos to offer lab testing in its stores (read more about Theranos here [Wired article about Theranos CEO Elizabeth Holmes] and here [NewYorker profile of Holmes]) and its "Balance Rewards for Healthy Change." I particularly enjoyed the short discussion about Walgreens' implementation of private exchanges. 

David Notari CEO of Innovation Health, a new insurance company that emerged from a collaboration between Aetna and Inova in the Northern Virginia market, also at the Monday Afternoon keynote, also provided a valuable perspective on the market. Innovation Health's plan is to leverage real-time information, for instance about patients having gone to the ER (instead of learning about it when they receive the claims), to deliver superior performance through an enhanced team-based care model. They also hope to use a full menu of network options to achieve optimum performance, since a key point of competition between plans (and payers) is the size of the network.

From #nhpc15: HSR Impact Award to Johns Hopkins for ACG Case Mix

The Health Services Research Award was given this year at the National Health Policy conference to Johns Hopkins researchers. Their work lies at the core of current case-mix assessment and risk adjustment. The overall idea is that an assessment of, say, providers' interventions or hospitals' performance must take into account the underlying health status of the population they serve. The methodology has become known as Ambulatory Care Groups or Adjusted Clinical Groups, both with the acronym ACG.

The award summary explains: "The original ICD-diagnosis code based epidemiologic approach for morbidity clustering now includes a wide range of risk, health status and disease measures based on diagnosis, medications, and other input factors derived from insurance claims, encounter/ discharge data and electronic medical records. The ACG software... also incorporates an array of comprehensive predictive modeling algorithms relevant to ambulatory care, hospitalization, and pharmaceutical use." It goes on to describing, at a high level, some of the uses for ACGs. For instance, "ACGs are key to setting risk adjusted capitation payments, global budgets, or incentive payments amounting to tens of billions of dollars annually." In particular, the ACG system has been successfully used for finance/budgeting applications to actuarial underwriting, capitation and resource allocation.

Here is the news release by AcademyHealth, which administers the award.

The Johns Hopkins website on the ACG software can be found here.

Hear Dr. Jonathan Weiner talk about ACGs and risk adjustment in the video below.

Specialty Pharmaceuticals in Health Affairs, Part 1

10.cover The October issue of Health Affairs is on speciality pharmaceuticals - those expensive prescription drugs that often find themselves in the news due to the financial burden they put on patients and payers while offering the promise of healing severe medical conditions. Here are a few articles that caught my attention. (Also, given the topic of this post, maybe my readers will be interested in a whitepaper I wrote about the Minimum Viable Market Share (MVMS) of stratified medicine in various market settings, to extend work by Mark Trusheim and co-authors in Personalized Medicine. Even if you don't like math, the tables should be relatively easy to understand since I've color-coded the MVMS based on how easy or difficult it would be for a pharmaceutical company to reach that threshold in market share.)

The impact of speciality pharmaceuticals as drivers of health care costs (B. Hirsch, S. Balu, K. Schulman, all from Duke). Excerpt: "Currently, 86 percent of prescriptions in the US market for small-molecule agents are for generic medications. This is an astonishing change from 1995, when only 40 percent of retail prescriptions were for generic medications. The industry therefore sees speciality pharmaceuticals as a way to offset losses from brand-name small molecules' losing patent protection." This paper is very good and very interesting for many reasons. Below, I'll mostly talk about the very basic mathematical model and Health Affairs's policy of relegating even the simplest math statement in its dreadful online appendix system, but it is only a small part of the paper. If you don't care about that, you can skip to the paragraph that starts with the header: "Going back to the paper".

The authors describe a simple economic model they created to evaluate the impact of specialty pharmaceuticals. In their words: "Let us assume that every person covered by a hypothetical insurance plan had a yearly out-of-pocket medical expense of $3,500 to cover his or her premiums, absent the use of any specialty pharmaceutical. The derivation of this value is presented in the online Appendix, as are the underlying assumptions of the cost model."

What you have to understand about Health Affairs (which is the leading journal in health policy that everyone dreams to publish in) is that it puts everything that remotely smacks of data or (always basic) models in an appendix, so that the paper itself only contains the high-level setup of the study, some tables and figures, attempts at analysis, and conclusions. But the online Appendix is, per Health Affairs guidelines (not the authors' fault), a double-spaced Word file formatted in 12pt Courier font, aka typewriter font, and if you've never seen what it looks like you really have to give it a try right now to get an idea of the ugliness of it. When you see a file formatted like that, you really don't want to read it. Health Affairs's idea of data is usually R-square values for regression or some really basic statistic concept that everybody in the health policy sphere seems to agree shouldn't be in the main paper, thus making it harder to discuss the authors' basic models once they have been green-lighted for publication.

Keeping the quantitative details of a study away from the main paper and in a format that only the most dedicated readers like yours truly will want to put up with helps feed the idea to non-quantitative policy-makers that anything about math is better left unseen (and frankly, as much as I do love Health Affairs and have learned a lot from reading it, which I can't say of any other journal, the mathematical models in the journal are extremely simple, usually focused on data analysis, sometimes using regression tools. The hardest part always seems to get your hands on the data. This is not to say that I don't like the authors' models, in fact I find them very interesting. But I think Health Affairs should make its policy to include more details about the models in the paper itself to foster discussions within the community. If this sounds anathema to policy-makers, then maybe the abstract can be extended for their sake.

In the present case, let's first talk about the $3,500 value, which the authors justify in the online appendix rather than the paper itself. It turns out they took a number from the 2013 Kaiser Family Foundation survey of $16,351 per family, assumed an average family size of 4 (which, I think, is an overestimate, since the US Census puts the average family size at 2.58, but maybe the authors assume insured people are more likely to have children? they never explain that), and a medical loss ratio of 0.85. We have 16,351*0.85/4=3,474 rounded up to $3,500. (And if they had picked 2.58, it would have been almost $5,400.) When you reach a situation where the authors feel compelled somehow to put 16,351*0.85/4=3,474 (which technically they don't really put, they just give the 16,351 and the 0.85 and the 4 separately) in an online appendix, you know you have a long way to go before data is accepted by policymakers, although this is not the authors' fault.

The authors assume that cost to payers and patients is $100,000 per patient and enrollment drops by 1% for every 10% increase in the cost. (It would have been good if they could have justified this number, which they never do.) The statement that "health care costs would be expected to increase by $250 for every o.25 percent of the population using the speciality pharmaceutical or $1,000 for every 1 percent increase in utilization" follows from the fact that the costs would increase by $100,000 times the prevalence in the population. If you inject prevalence = 0.25% you get $250 in cost increase or 7.14% compared to the initial $3,500 amount, although the authors curiously put the percentage increase at 6%. (This may be a typo, because the numbers for 1% prevalence do amount to a 29% cost increase as stated.) And, that's it for the online appendix!

Because there is a 29% cost increase at 1% prevalence, enrollment drops by 2.9% given the authors' assumption of 1% enrollment drop per 10% cost increase. This means that premiums must increase by 2.9% to cover the enrollment drop. In the article, the authors then show Exhibit 2 (Rate and Percent Increase in Insurance Premiums for a New Specialty Drug Costing $100,000 per Treated Patient Depending on Disease Prevalence.) So while the authors had warned their model was "intentionally simplistic", the online appendix ended up a little underwhelming, to say the least.

Again, that's not a comment on the authors, but on Health Affairs's philosophy. (Basic models can provide very valuable insights into a topic. There's no need to go right away to a complex model if a basic model can easily give us the big picture.) Realistically, the math involved is really very basic multiplication and division of numbers (not unknowns or parameters) of the type you learn in, what, middle school. That sort of math should not be relegated to an online appendix in Courier font that doesn't even look like it belongs to the same journal, not being typeset in the journal's template.

And what happens if we use the average family size of 2.58? $1,000 is then only 19% of the baseline cost, leading to a premium increase of 1.9% or about one third lower than what the authors state. But since the authors forgot to reference their "1% dropout per 10% cost increase" number, it doesn't mean the 1.9% premium increase is any more meaningful than the 2.9% in Exhibit 2 of the paper.

Going back to the paper

The authors also point out that those increased costs could be mitigated by decreased hospital admissions. They then investigate cost-sharing through copayments and co-insurance and discuss the influence of section 340B of the Veterans Health Care Act of 1992. They explain: "The purpose of the [340B] program is to ensure access to pharmaceutical agents at safety-net hospitals." Following the Omnibus Budget Reconciliation Act (OBRA) of 1990, "pharmaceutical companies were less willing to provide aggressive discounts to safety-net hospitals" because the prices those hospitals received were now used to establish benchmarks, for the rebates that OBRA required pharmaceutical companies to offer state Medicaid programs. The 340B program ensured that "prices under this program would not be considered in establishing Medicaid rebates under OBRA" for hospitals providing high levels of uncompensated care. This has led to 30-50 percent discounts off the market price. There is thus an "enormous financial benefit" in being enrolled, and in fact, one-third of all US hospitals now participate in this program. 

Yet, the authors state, "The 340B regulations do not limit the application of discounts received by hospitals to medications used in the care of indigent patients, nor do they require hospitals to pass their cost savings along to payers or patients." The program is now expected to cost $12 billion by 2016 and may lead to manufacturers increasing their prices "to compensate for the loss in revenues related to the program." The authors provide convincing examples as to why this "may have negative economic effects on patient care."

Then they discuss biosimilars and price competition, especially given the expected loss of patent protection for several biologic agents in 2013-18. They observe that: "This approval pathway [for specialty pharmaceuticals] is not subject to the generic drug provisions of the Drug Price Competition and Patent Term Restoration Act of 1984... so there is not a standard pathway for competitors to enter the market once a specialty pharmaceutical's patent expires." They point out, however, that the Affordable Care Act "included provisions for the approval of biosimilar products" but also that "in 2010 the Congressional Budget Office estimated that biosimilars would yield only a 2 percent reduction in pharmaceutical costs by 2019."

If there is only one message to keep from this well-written and well-researched paper, it is that specialty pharmaceuticals present valuable opportunities for new innovation pathways but there is an urgent need for far-ranging discussions on their implications in the health care market.

The Finance Side of the Healthcare System: Links

For today's post, here are a few resources about health policy that I found interesting.

Affordable Care Organizations

A December 2006 Health Affairs article by Dartmouth's Elliott Fisher et al, Creating Accountable Care Organizations: The Extended Hospital Medical Staff, suggests ACOs should consist of local hospitals and the physicians who work around them. The authors observe that "virtually all physicians are either directly or indirectly affiliated with a local acute care hospital, whether through their own inpatient work or through the care patterns of the patients they serve." This allows them to "empirically define the multispecialty group practices that [they] refer to as an 'extended hospital medical staff.' " 

A January 2011 article by Jeff Goldsmith of the University of Virginia, Accountable Care Organizations: The Case for Flexible Partnerships between Health Plans and Providers, makes the case that "health care services [should be divided] into three categories: long-term, low-intensity primary care; unscheduled care, including unscheduled emergency services; and major clinical interventions that usually involve hospitalization or organized outpatient care. Each category of care would be paid for differently, with each containing different elements of financial risk for the providers."

A January 2014 article by Arnold Epstein of Harvard School of Public Health et. al., Analysis of Early Accountable Care Organizations Defines Patient, Structural, Cost and Quality-of-Care Characteristics, analyzes the profile of ACO enrollees and compares it to non-ACO enrollees in order to establish a baseline for further program assessment.

ACOs have also been the topic of a number of Health Affairs blog posts, such as this one (payment reform landscape), this one (a look ahead for 2014: fantastic resource about the growth in ACOs, lives covered and geographic dispersion), this one (on current data about which ACOs, whether Pioneer, Medicare Shared Savings Program, Commercial or Medicaid, were financial winners or losers) and that one (about the Totally Accountable Care Organization, with the very unfortunate acronym TACO, although the authors themselves use that acronym and thus must be rather happy with it - maybe someone just has to come up with something healthcare-related that would make the acronym BELL and the health care system will be saved! Side note: in French, a tacot - you don't pronounce the T - is an old, beat-up vehicle, basically a clunker. I swear I'm not making it up. The authors' TACOs refer to Medicaid ACOs. They explain that: "“Totally” refers to the expectation that these organizations will be responsible for services beyond just medical care (for example, mental health, substance abuse treatment and other social supports), as well as the aspiration that these organizations will assume accountability for all associated costs of care, ultimately, through global payment mechanisms.")


From a 2013 Kaiser Health News article comes a description of the ACA's Reinsurance Tax, which charges each health plan $63 per enrollee to help cover 80% of the costs for claims between $60,000 and $250,000 per person. According to the Department of Health and Human Services, premiums for individual coverage would be about 10-15% higher without the reinsurance tax (see link in the KHN article). Interestingly, it seems from the KHN article that the tax is also assessed on enrollees of employer-sponsored plans, although those employers and their health payers cannot access the tax's benefits. 

A 2005 report from the Commonwealth Fund provides a good overview of reinsurance and how it can help states lower premiums. If you don't have time to read the PDF, you can always read the issue brief instead. At the time the brief was written, state-level reinsurance programs had only been implemented by New York (excess-of-loss) and Arizona (aggregate stop-loss). Being the analytical nerd that I am, I found the details of those two reinsurance programs just fascinating. First, definitions: as you might have guessed, New York focuses on reinsurance for individual enrollees, i.e., it protects insurers "for the risk of extraordinarily high costs incurred by any individual." In contrast, the Arizona program "provides protection to insurers for the risk that a large number of enrollees may have above-average but not necessarily extraordinary expenses—a situation that typically occurs when they are more likely to have chronic health problems."

The programs also offer different incentive structures for insurers. In New York, "the insurer is responsible for 10 percent of the costs between $5,000 and $75,000 and all of the costs above $75,000." In Arizona, "the plan encourages insurers to reduce total costs."

The report also has data on how reinsurance lowered premiums. For instance, "In New York City in 2004, Healthy New York premiums were 40 percent lower than the average small group HMO premium and two-thirds lower than the self-pay individual market premium, while in Arizona, "the total state subsidy was $8 million per year [at the inception of the program in FY2001]. By FY2004, because expenses had been significantly reduced, this subsidy could be cut to $4 million."

There is also an interesting 1997 report on reinsurance prepared for HHS, but it is so old that the numbers provided in the analytics part have become completely outdated. 

Narrow Networks

This Kaiser Health News article aggregates several news articles about trends in 2015 regarding narrow networks. A great reference if you're curious to learn more about what's going on.

Using Big Data To Transform Healthcare

7.cover-2The July 2014 issue of Health Affairs is on "Using Big Data To Transform Care." Here are some papers that caught my attention.

Big Data in Health Care: Using Analytics To Identify and Manage High-Risk and High-Cost Patients, by D. Bates, S. Saria, L. Ohno-Machado, A. Shah and G. Escobar. (Paper) This article presents, at a high level, six use cases "where some of the clearest opportunities exist to reduce costs through the use of big data: high-cost patients, readmissions, triage, decompensation (when a patient's condition worsens), adverse events, and treatment optimization for diseases affecting multiple organ systems."

Optum Labs: Building a Novel Node in the Learning Health Care System, by P. Wallace, N. Shah, T. Dennen, P. Bleicher and W. Crown. (Paper) This paper is about Optum Labs, a joint entity created in early 2013 between the Mayo Clinic and Optum, an organization part of UnitedHealth Group. Partners now include AARP, AMGA (the American Medical Group Association), Boston Scientific, Boston University School of Public Health, Lehigh Valley Health Network, Mayo Clinic, Optum, Pfizer, Rensselaer, Tufts Medical Center and the University of Minnesota School of Nursing.

Here are examples of research findings:

  • There are a number of seemingly equivalent choices for second-line medication treatment for uncomplicated type 2 diabetes. (Second-line medication is medication you give if the patient doesn't respond as expected to first-choice or first-line medication.) An analysis of the Optum data, however, showed that a specific drug led to less cost and a longer interval until insulin was required. This has the potential to change guidelines used by care providers.
  • Another Optum study showed the number of total knee replacement surgeries has increased steadily for all age groups, as has the prevalence of diabetes and obesity among patients undergoing these surgeries. Hence, it seems important in future studies to differentiate patients with multiple comorbidities.
  • Another Optum study showed that "the statistical linkage of clinical information from laboratory test results to claims data reduced bias in treatment effects estimates by 24 percent for all ambulatory visits and by 79 percent for visits related to Hepatitis C, relative to the use of claims data alone."

A key overall challenge (and priority) remains to "enable clinicians to connect findings from big-data analyses directly to the care of an individual patient".

Leveraging the Big-Data Revolution: CMS is Expanding Capabilities to Spur Health System Transformation, by N. Brennan, A. Oelschlaeger, C. Cox and Marilyn Tavenner. (Paper) CMS has collected claims, assessments and surveys for many years, and has long made them available to researchers. However, data was previously shipped to researchers in the form of encrypted hard drives, at a rather significant cost for the researchers. In late 2013, the agency launched the CMS Virtual Research Data Center. Researchers can now access the data from their own computers but they "may download only aggregated data sets and analytic results." This is obviously expected to decrease the cost to the researchers (no more encrypted hard drives arrivign through the mail!) Innovations have also been made in the qualified entity program to improve provider quality reporting and in sharing data with providers and states. Further, the Blue Button program "allows patients to easily access their own health information in electronic form."

Early Experiences with Big Data at an Academic Medical Center, by J. Halamka. (Paper) This describes the experience of Beth Israel Deaconess Medical Center (BIDMC) in the Boston area in adopting electronic applications. While data storage now amounts to 3 petabytes (1 PB = 10^(15) bytes) and grows at a rate of 25% a year, the main issue has shifted to the transformation of this raw data into usable information.

In order to gain insights, BIDMC created "Clinical Query, a web-based tool that allows for analysis and visualization of the clinical data collected by BIDMC." The author explains: "The principle behind Clinical Query is that investigators will want to ask questions before conducting research that will help them understand the potential statistical power of a clinical trial or the availability of data for clinical research." The resulting web-based query tool uses Boolean (and/or) expressions and open-source code from the Harvard's Integrating Information from Bench to Bedside project. The author identifies the potential issues related to Big Data: (1) data quality, (2) variations in data collection over time, (3) inconsistent use of medical terminology, (4) patient privacy concerns, (5) need for experts to meaningfully analyze data.

This issue of Health Affairs was very interesting. I do wish that an InnoCentive-type competition where de-identified claims and diagnostics data would be posted online (perhaps using the CMS virtual system with its built-in protections) for all interested parties to study and analyze. When the data is only accessible to some groups of researchers, opportunities for insights are lost because other teams of researchers may have approached the problem a different way or tested a different hypothesis. Further, I also think that the field is ill-served by the current lack of mathematical models on the topic. Health Affairs never presents mathematical models to begin with, but when it comes to analytics, there comes a point where the buzzwords have to stop and the analytical insights have to begin - and those analytical insights have to extend beyond the data-mining insights that are already being derived. (Data-mining is a good first step, but operations research can do so much more.) Given the tremendous progress that has been made in a few short years, though, the state of analytics applied to healthcare treatments is sure to become quite impressive indeed.

Robust risk adjustment in health insurance

This is a short paper my doctoral student Tengjiao Xiao and I recently completed. 

Here is the abstract: "Risk adjustment is used to calibrate payments to health plans based on the relative health status of insured populations and helps keep the health insurance market competitive. Current risk adjustment models use parameter estimates obtained via regression and are thus subject to estimation error. This paper discusses the impact of parameter uncertainty on risk scoring, and presents an approach to create robust risk scores to incorporate ambiguity and uncertainty in the risk adjustment model. This approach is highly tractable since it involves solving a series of linear programming problems."

The paper also contains, in the section where we motivate the need for robustness, the graph about ranking changes using proxy and actual Value-Based Purchasing factors that are used to give the about 3,000 hospitals considered bonuses or penalties. A negative ranking change indicates a loss in ranks and a positive one indicates a gain. The interesting thing about this graph is that losses and gains can fluctuate enormously, meaning that some hospitals that would have stood to receive very high bonuses (for the amounts of money considered: every hospital contributes 1% to fund the scheme) under proxy factors found themselves at the very bottom of the ranking, and vice-versa. To the best of our knowledge, this is not something that has received much if any attention in the media. 

The core of the short paper is to show how robust risk scores can be computed by solving a series of linear programming problems, with the aim of minimizing worst-case regret between the actual risk scores, used to implement transfer payments between health payers, and the true scores, which we don't know. We show on a simple test case with 10 insurers that the change in payments can be substantial.

Comments welcome!