Imran Cronk

This article was originally published in Harvard Business Review and co-authored with Kartik Hosanagar, a professor at the Wharton School.

On May 7th, 2016, Joshua Brown, a 40-year-old entrepreneur and technology enthusiast from Canton, Ohio, was sitting behind the wheel of his Tesla Model S sedan when a tractor-trailer turned across his path. The Tesla, which was engaged in its self-driving Autopilot mode, failed to register the white tractor-trailer against the bright, sunny Florida sky. Mr. Brown also didn’t engage the brakes in time. His Tesla crashed into the truck at 74 miles per hour, killing him almost instantly.

More than 30,000 people are killed each year in car crashes in the United States. In 90% of crashes, human error is to blame. And so most experts agree that self-driving car technology will reduce the number of crashes and fatalities. Self-driving cars, Adrienne LaFrance writes in The Atlantic, could save up to 1.5 million lives just in the United States and close to 50 million lives globally in the next 50 years. Yet in a March 2016 poll by the American Automobile Association, 75% of respondents said they are not ready to embrace self-driving cars.

Driving a car is one of the most personal – and dangerous – things we do. It’s understandable that people are skeptical of handing over their keys to a faceless algorithm and sitting back for the ride. When you think of the word “algorithm,” you might picture a computer crunching numbers according to a formula or following a pre-programmed sequence of steps. But algorithms have come a long way in the last decade: they can take in data, learn, and generate more sophisticated versions of themselves. They can even drive a car.

We rely upon algorithms for many of our decisions and actions, from low-risk activities such as deciding what to watch on Netflix or buy on Amazon to high-stakes decisions such as how we should invest our savings. We are even OK with autopilot features controlling our airplanes. This current skepticism for self-driving cars thus raises a question: Why do we trust algorithms in some cases, but not in others?

Our Selective Trust in Algorithms

Humans aren’t always algorithmically averse.  Research conducted by one of us (Kartik) on automated product recommendation algorithms, such as Amazon’s “People Who Bought X also bought Y,” found that people like algorithmic recommendations and often follow their advice. For example, in one recent study conducted with professor Dokyun Lee at Carnegie Mellon University, we randomly assigned consumers at a top-five online retailer in Canada to either a treated group that received algorithmic recommendations or a control group that received no recommendations. We found that the algorithmic recommendations drove a 25% increase in the number of products viewed by consumers and a 35% increase in products purchased. In additional research, we found that the influence of recommendation algorithms on choices is greater for hedonic products – characterized by pleasure-oriented consumption (e.g., movies, perfume, art pieces) – than for utilitarian products wherein consumption is motivated by functional need (e.g., paper clips, dishwashing agents and vacuum cleaners).

This paper was originally published in a scientific journal called the Annals of the New York Academy of Sciences and co-authored with John Kimberly, a professor at the Wharton School. Just the abstract is below, but you can find the full article here.

The world of health care is changing dramatically, as reflected in the number, magnitude, and scope of innovative new approaches—to how illness is treated and how better health is promoted—that are being implemented around the globe. The changes triggered by these initiatives affect both how care is organized, managed, and paid for and the kinds of approaches that are being developed to keep people healthy. Underlying these changes is a more fundamental paradigm shift, a shift in the priority given to “value” in the formulation of policy and management practice. This brief essay highlights five trends that are central in this shift: increasing emphasis on health promotion, movement toward value-based payment, advances in digital/mobile technology, exploitation of big data, and changes in support for biomedical research. Each of these has its own value controversies, and the individual impact of each is impossible to predict. Collectively, however, their impact is likely to be significant. [Full Article]

This post was originally published for the Health Policy$ense blog of the Leonard Davis Institute of Health Economics.

Dr. Grace Terrell, President and CEO of Cornerstone Health Care, recently visited LDI and shared her experience of directing a physician-led health system through health care reform. Terrell, a primary care physician and a good Southern storyteller, told us about ‘Julia’, her patient of more than 20 years:

Julia had just gotten a huge medical bill [from] this place in the local community that could cure all sorts of knee pain and back pain. What she got was a $1,500 bill that her exchange product paid about 60 percent of, for care she absolutely did not need. It was unnecessary, expensive, inappropriate care at the wrong place, for the wrong price, at the wrong time. Unfortunately, that’s the way a lot of health care is still in this country. The story of our organization is confronting that and dealing with that and trying to innovate around that in the middle of being in just a regular old medical practice.

Cornerstone began in 1995 as a multidisciplinary group of 42 physicians in 15 practices in central North Carolina. It focused on early adoption of new technology and practice innovations, including: electronic medical records in 2005; weekend hours in consolidated, multi-specialty facilities in 2007; and certification of its primary care practices as so-called “medical homes” in 2008. In 2015, Cornerstone had a high national rank among Medicare Shared Savings Program participants, and its spinoff CHESS has been selected to become a Next Generation ACO.  More than 300 physicians are now part of the group.

Gauging from her storytelling, Terrell is clearly passionate about designing new models of care to deliver greater value in health care. Some examples of Cornerstone’s initiatives:

• Cornerstone invested in population health analytics software and reached out to patients who had lost touch with their doctor. According to Terrell, Cornerstone had “better results with blood pressure, cholesterol and blood sugar from just this one maneuver than we had from hiring our new endocrinologist at $300,000 a year!”

• In a “look at the whole picture approach”, Cornerstone set up two clinics with an internist, a nurse practitioner, a care navigator, behavioral medicine specialist and a pharmacist.

• Cornerstone established a heart function clinic with embedded behavioral services since “the number one indicator for heart failure readmission to the hospital is actually depression.”

• In its oncology clinic, Cornerstone embedded a general internist who could preserve continuity of care for patients’ medical needs beyond cancer.

• For Medicare and Medicaid dual eligibles, Cornerstone created a concierge practice with full assessment of psychiatric needs.

• To support the neediest and sickest patients, Cornerstone built an extensivist practice with a focus on medication management.

Unstable Finances
However, the outcomes they achieved did not translate into a stable financial base. Terrell observes: “It has been an up-and-down, yin-yang experience for our organization where the finances have never been there as we had thought they were going to be…The payers have never been as quick to move as we thought they were going to move.”

For much of the past year, Cornerstone has been “trying to keep the place open and pay the lights and doing things like pay chemo bills and things like that…It required a significant amount of working with the physicians who said ‘we thought this was going to work by now.’” In early 2016, Cornerstone was acquired as a wholly-owned subsidiary of Wake Forest Baptist Medical Center, an academic medical center based in Winston-Salem, NC. It will continue to operate as a separate business unit.

Looking ahead, Terrell ties up her story about ‘Julia’: “I don’t know where the rest of things are going, but I do know it’s the right thing to do. We’ve got an incredible problem in this country: a sixth of our economy is giving health care to one another. We don’t invest in anything else much, and we don’t give good care – or we give care like we did to Julia. The $1,500 got spent on something of absolutely no value, when she’s having all sorts of other medical issues that are not being addressed because we haven’t had the infrastructure to do it.”

This post was originally published for the Health Policy$ense blog of the Leonard Davis Institute of Health Economics.

Penn’s Second Annual Martin Luther King, Jr. Health Equity Symposium drew attention to the importance of inclusion and diversity in medical education and research on both a national and local level.  A panel of Penn faculty, including several LDI Senior Fellows, directly confronted the barriers to inclusion at Penn, and Dr. Richard Carmona, 17th Surgeon General of the United States, shared a detailed account of his rise from an impoverished Hispanic family in Harlem, New York to the prestigious post as the “doctor for the nation” in the Bush administration.

Dr. Eve J. Higginbotham, Vice Dean for Inclusion and Diversity, introduced the panel, and Dr. Jerry Johnson, Chief of Geriatric Medicine and Director of the Center of Excellence for Diversity in Health Education and Research moderated the group. Each panelist gave her perspective on the importance of diversity, barriers to progress, and possible solutions.

• Dr. Tiffani J. Johnson from the Children’s Hospital of Philadelphia explained how the “leaky pipeline” prevents underrepresented minority students from advancing through the ranks of medical school to residency to junior and tenured faculty. Part of the issue, she said, is our implicit bias against racial minorities that lies “below the surface, but may influence behavior.” While racial bias is well-documented in the business and academic worlds, Johnson shared evidence that pro-white and anti-black racial bias exists among physicians for both adult and pediatric patients. She suggested that implicit racial bias could be mitigated through “positive black priming” and increasing interactions between people of different races.

• Dr. Jaya Aysola from the Perelman School of Medicine discussed the factors contributing to a culture of inclusion (and lack of inclusion) at Penn Medicine. Her research has found significant variation in experience according to gender, ethnicity, and sexual orientation. In particular, women, LGBTQ, black, Hispanic, and multi-ethnic individuals perceive a lower “cultural competence” at Penn. Aysola called for identifying and improving factors within Penn’s organizational system and culture.

• Dr. C. Neill Epperson from the Penn Center for Research on Sex and Gender in Health challenged the audience to think about diversity and inclusion among medical and health services researchers. She shared data indicating that low institutional support, low values alignment, low inclusion and low self-efficacy made people more likely to leave their institution – and that underrepresented minorities experience these issues at high rates. To address the barriers faced by younger researchers, she pointed to solutions such as on-site daycare and more progressive parental leave policies.

• Dr. Shreya Kangovi, founding executive director of thePenn Center for Community Health Workers explained how the community health worker model can improve access and quality of care, improve patient activation and mental health, and reduce readmissions. The Penn CHW center delivers care to 1,500 patients per year and has advised more than 500 organizations who also want to develop a program. Kangovi related a story that captured the value of using CHWs as preceptors for medical students in low-resource environments: “30-year old, no family, uninsured and taking street Xanax. You automatically think: difficult patient. We walked in and [CHW] was like oh my god, your hair is so cute! The patient got this big smile on her face and started talking to us. My whole impression of her just changed.”

In his closing remarks, Dr. Richard H. Carmona talked about his journey from a poor Hispanic family in Harlem, New York to the U.S. Surgeon General position in the Bush administration from 2002-2006. His tenure is notable for hislandmark report on the harms of secondhand smoke and his subsequent criticism of the Bush administration for suppressing his public communications related to stem cell research, contraception and climate change.

Dr. Carmona delivered a passionate call for greater efforts to reduce inequities in health. After recounting “sobering” statistics about health inequities between blacks and whites—including how black women are 2.5 times more likely to die during pregnancy—he discussed how racial inequities pervaded every aspect of his agenda as Surgeon General. During his term, he grew to believe that the issue was one that we could neither ignore nor escape from.

“Martin Luther King recognized these injustices and inequities,” said Carmona. “He understood the social determinants of health — how all of these things lead to bad outcomes. When people don’t have access, when they don’t understand, what they can’t make informed decisions on what they need to pursue optimal health and wellness.” Carmona also called out Congress for delaying progress.

 Congress remains divided and fights over this because, ‘Well, we don’t want another welfare program.’ Well, neither do I. I want to empower people. […] If we don’t do something about these disparities, injustices and so on, the disease and economic burden we will leave our children is unsustainable.

Whether you have a heart, or whether you’re just a smart businessman, we have compelling reasons to start interceding aggressively to eradicate these disparities.

The audience responded with a standing ovation at the end of his talk. You can hear a clip here.

The Third Annual Martin Luther King, Jr. Health Equity Symposium will take place on Monday, January 25, 2017 and will feature a keynote address from Dr. Antonia Novello, the fourteenth Surgeon General of the United States, who served from 1990-1993.

This post was originally published for the Health Policy$ense blog of the Leonard Davis Institute of Health Economics.

Despite decades of calls for increased representation of minorities in the health professions workforce, we are very far away from a workforce that reflects this nation’s diversity. Underrepresented minorities make up 31% of the general population, but just 15% of medical school students and 13% of dental students. A new study helps us understand the barriers minority college students face in pursuing medical and dental careers.

In Academic Medicine, Brandi Freeman and colleagues, including LDI Senior Fellows Judy Shea and David Grande, report on focus groups they conducted with undergraduates from minority backgrounds that are underrepresented in medicine, including Blacks, Latinos, and Native Americans. The one-hour focus groups, involving 82 diverse students across 11 colleges, highlighted several challenges: inadequate institutional resources for academic success and clinical opportunities; strained personal resources such as lack of financial resources or familial pressure; inadequate guidance and mentoring to assist with key career decisions; and societal barriers such as work-life balance concerns or job uncertainty.

The quotes from the focus groups illustrate the challenges, insecurities and uncertainties that these students face:

…somebody else who never worked—had to work for anything and their parents paid for all their college, it’s their GPA is obviously going to be higher because all they had to focus on was school.

It’s kind of a disadvantage almost if you don’t have family that—or someone that will let you come into their workplace and follow them around. […] For people who don’t have that as an option, it makes us look bad.

I feel kind of lost. I know I want to be there, but I just don’t know how to get there.

What happens if you never get matched, I guess? Because that’s a possibility and you don’t go through residency, so you’re stuck with an MD who can’t practice medicine.

The focus groups were conducted as part of the Tour for Diversity in Medicine, an effort from underrepresented minority physicians and dentists to encourage students of diverse backgrounds to pursue careers in the health professions. The focus group approach to understanding root causes of the “leaky pipeline” is important, the authors say, since past studies have relied on quantitative data such as academic achievement, focused on a single institution, or captured perspectives from minority students who have already become health professionals. This new study is more qualitative, involves multiple institutions across the nation, and captures the undergraduate perspective.

The authors suggest that external programs, such as the Summer Medical Education Program(SMDEP), can strengthen support for students at resource-limited institutions. To address strained personal resources and familial barriers, the authors recommend educating families at the high school level to familiarize them with the medical training process at an earlier stage. At a higher level, the authors suggest that policy changes, such as regulation of medical resident work hours, can change perceptions of work-life balance.

Increasing the diversity of the workforce is important because health professionals from underrepresented backgrounds disproportionately serve minority and other medically underserved populations. In addition, minority patients tend to receive better care from practitioners of their own race or ethnicity, particularly in primary care and mental health settings. As part of a series on how the Affordable Care Act affected minority health last January, we wrote about workforce diversity. The ACA invested $100 million to expand scholarships and loan repayments for disadvantaged and minority students; provided large grants to historically Black Colleges and Universities for academic support, faculty development, and research surrounding health issues; and created $67 million in Health Profession Opportunity Grants (HPOG) for low-income families.

The authors of the focus group study point out that the perceived and actual barriers for minority students in the health professions pipeline are similar to those for other science degrees and fields, and that interventions can affect diversity across a broader set of careers. One closely related field is health services research, which the Leonard Davis Institute and other Penn institutions support through the Summer Undergraduate Minority Research (SUMR) program. Now in its 16^th year, SUMR provides stipends for students to conduct research with Penn faculty on a topic of their choice. These programs are an important step in addressing barriers for minorities who want to help advance the nation’s health.

This post was originally published for the Health Policy$ense blog of the Leonard Davis Institute of Health Economics.

While the Affordable Care Act has achieved a second victory before the Supreme Court and produced significant coverage gains, it might also have produced a less positive outcome: in an NBER working paper, Penn LDI colleagues Mark Pauly, Adam Leive and Scott Harrington found that a large portion of non-poor (measured by income above 138% of the poverty level) who gained coverage now have a higher financial burden and lower welfare (well-being) than when they were uninsured. The authors call this extra burden a “price of responsibility” for complying with the individual mandate to purchase coverage.

To evaluate the change in financial burden and welfare, the authors compared the out-of-pocket payments made by uninsured people before the ACA with premiums and out-of-pocket payments made after gaining coverage. The authors also estimated the positive effects of health coverage, such as higher use of services and protection from catastrophic medical bills. Even so, the model found that non-poor adults who went from uninsured to insured were paying higher premiums (even with subsidies) and, surprisingly, more out-of-pocket fees. While the burden was lower for those with lower incomes, because of subsidies for premiums and co-pays, the burden across all levels of income was positive – meaning that the average non-poor adult who gained insurance under the ACA had a higher financial burden after purchasing insurance.

The authors estimated that subsidy-eligible people with incomes below 250% of the poverty threshold likely experience welfare improvements that offset the higher financial burden, depending on assumptions about risk aversion and the value of consuming more medical care. However, even under the most optimistic assumptions, close to half of the formerly uninsured (especially those with higher incomes) experience both higher financial burden and lower estimated welfare.

Stated succinctly:

“Persons with low incomes may fare better after the ACA, but those formerly uninsured at higher incomes not in poor health consistently are worse off.”

The implication here is that middle class people with low perceived health risk might prefer to remain uninsured and pay the penalty for violating the individual mandate. Reluctance among healthier and higher-income uninsureds is no surprise, but this paper appears to be the first to robustly measure the actual trade-off they would have to make in purchasing insurance.

Given that insured people use more than twice as much health care as uninsured people, it is not so hard to imagine that formerly uninsured people now have more responsibility for premiums and co-pays. But how did the authors conclude that the upsides of insurance – risk protection and actual services – are not enough to outweigh the financial burden and create “positive welfare” for newly insured people?

One reason is that most of the formerly uninsured were receiving some amount of free care (“bad debt” or “charity care”) before the individual mandate took effect. What’s changed is that people are on the hook for premiums and co-pays to receive that same care, plus other services that might not have been provided for free. The authors acknowledge that, for individuals who are low-income or at a high risk for expensive care, purchasing insurance can lead to improved welfare from additional health care services. Looking out across the entire group of uninsureds, though, the benefit of additional health care services does not appear to outweigh the increased financial burden, even with subsidies.

The surprising findings have sparked reaction from across the economic blogosphere. Matthew Martin from Separating Hyperplanes proposes one explanation: “the ACA is especially goofy in that much of the redistribution is confined to within the new individual market – even though people with employer-sponsored coverage are generally both wealthier and healthier. We have community rating within large employers and within the new individual market, but there’s no mechanism to redistribute between each of these pools.”

Writes Tyler Cowen from the blog Marginal Revolution: “I’ve read so many blog posts taking victory laps on Obamacare, but surely something is wrong when our most scientific study of the question rather effortlessly coughs up phrases such as […] ‘Average welfare for the uninsured population would be estimated to decline after the ACA if all members of that population obtained coverage.”

He continues, “the best thing to do is to improve it from within. Still, there are good reasons why it will never be so incredibly popular.”

For their part, Pauly, Leive and Harrington conclude: “It will be important to examine the level and pattern of these increased financial burdens to judge whether they are of sufficient social value to justify their imposition.”

This post was originally published for the Health Policy$ense blog of the Leonard Davis Institute of Health Economics.

Enhancing access to primary care is a key component of a patient-centered medical home (PCMH). But little is known about how PCMH status affects the availability of appointments for new patients. In a new analysis of “secret shopper” data, LDI Senior Fellows Jaya Aysola, Karin Rhodes and Daniel Polsky found that PCMHs were 1.26 times more likely to offer a new appointment and 1.36 times more likely to schedule an after-hours appointment than other primary care practices, with no differences in average wait time for a new appointment.

The data were collected in 2012-2013, prior to full implementation of the Affordable Care Act. Trained field staff placed more than 11,000 phone calls to more than 7,000 primary care practices across 10 states, posing as new patients seeking a primary care appointment. Previously, findings from the study showed differences in the likelihood of scheduling an appointment by type of insurance, and the important role played by federally-qualified health centers and rural health clinics in assuring appointment availability to Medicaid patients.

While just 5% of practices in the study were PCMHs, the difference in new appointment availability may take on increased importance as the model is more widely adopted and as millions of non-elderly adults gain coverage through the ACA.

Why might PCMHs have more new patient appointment availability than other practices? It could be simply because they make a concerted effort to make access to appointments easier, as part of their overall policies on enhanced access to care. Beyond this, the data are silent. Another hypothesis is that the efficiencies created by PCMH processes may allow for greater patient panel size. There is debate, however, on whether these efficiencies will instead be applied to improving the care for existing patients. The authors note:

Some believe that PCMH will expand panel sizes and assert that global payment schemes would naturally incentivize this over fee-for-service models. Others expect that PCMH practices will keep panel sizes low and increase the intensity of services provided to existing patients, by lengthening patient visit times to improve the quality of care and minimize provider burnout.

The study did find differences in average daily census per physician between PCMH practices and non-PCMH practices. Most physicians in both PCMH and non-PCMH practices saw an average of 20-­39 patients daily, but fewer PCMH providers saw more than 40 patients daily than those in other practices. However, the study found no significant relationship between average physician daily patient census and access to new appointments, and so the question remains an open one.

The PCMH model: what we know

Patient Centered Medical Homes (PCMH) are primary care practices that are accredited by the National Committee on Quality Assurance (NCQA) according to a set of standards that focus on enhancing access and continuity of care, identifying and managing patient populations, tracking and care coordination, providing effective care management, self-care and community support, and measuring and improving performance.

The PCMH model is still in its infancy; the NCQA proposed operational standards for recognizing practices as PCMHs in 2008. Given the time needed for “practice transformation,” and the wide variation in performance on the scale that the NCQA uses to evaluate PCMH-certified providers, comprehensive and reliable evaluations have been difficult to conduct.

A 2013 systematic review found evidence of a small positive effect on patient experiences and  delivery of preventive care services, but concluded that current evidence is insufficient to determine effects on clinical and most economic outcomes.

A recent study by Aysola and colleagues found that most patients enrolled in PCMHs within the University of Pennsylvania Health System didn’t even know that they were in a PCMH. Patients uniformly lacked awareness of the PCMH concept, and the vast majority perceived no PCMH-related structural changes, regardless of the degree of practice-reported PCMH adoption.

As the PCMH model spreads and evolves, and providers are able to move past meeting a list of standards and move toward meaningful transformation – in areas such as appointment flexibility, care coordination and remote support –more useful data on PCMH performance and outcomes should emerge.