Via Walid Gellad:
How infectious is that virus? How good is free birth control? How much should you worry about supplements?
For those of you who came here for references or more information, here you go:
- No, Seriously, How Contagious Is Ebola?
- Provision of No-Cost, Long-Acting Contraception and Teenage Pregnancy
- A synthetic stimulant never tested in humans, 1,3-dimethylbutylamine (DMBA), is identified in multiple dietary supplements
I’ve been travelling, so I didn’t get to post this until now. If you watch one thing on Ebola, make it this:
Last year, I wished TIE had a media award, so I could give it to Peggy Orenstein. I think Shep Smith has locked up the 2014 award with the above.
The following is a guest post by Jennifer Gilbert, a research assistant for Dr. Ashish Jha at the Harvard School of Public Health and for The Incidental Economist. She graduated from Boston University in 2014. You can follow her on Twitter: @jenmgilbert.
When people with visible disabilities pass by, they regularly attract the attention of others. Some well-meaning observers may imagine themselves in the same position and envision a world of sadness. But this is not the world that most people with chronic illness or disability inhabit.
Research shows that most people at least partially adapt to their condition. However, study design matters. The more optimistic studies assumed that having a high level of self-reported well-being equates to a return to baseline happiness, while less optimistic results emerged from research considering data before accidents, enabling analysis of life satisfaction over time.
One early, optimistic theory of adaptation stems from a 1978 study, “Lottery Winners and Accident Victims: Is Happiness Relative?” The investigators interviewed 29 paralyzed accident victims and 22 lottery winners, and found after one month to one year of time had elapsed, both groups returned to the level of happiness they’d had before winning the lottery or becoming injured. The authors coined the term “hedonic adaptation” for the idea that people have a general baseline level of happiness and will return in time to this set point as they navigate the ups and downs of life.
A large body of research has examined positive adaption to disability. These studies found that after developing a disability, self-reported happiness increases over time. However, the majority of them only measure level of happiness after onset of disability; they don’t use a baseline level of happiness for comparison.
According to one study of individuals in 11 social networks of persons with disabilities, a majority of the respondents with moderate to serious physical disabilities reported having an excellent or good quality of life. Another study found that 100 disabled participants reported levels of well-being that were only slightly lower than those of age-matched nondisabled people on scales of psychological well-being, life satisfaction, and depression. More recently, a 2005 study found that hemodialysis patients were no less satisfied with their lives than nondisabled people, and several studies have found paraplegics adapt in many ways, such as reporting improved mood within weeks of their accidents and experiencing similar happiness to others. These studies correlate higher levels of happiness with greater independence, supportive social networks, and specific personality traits.
It’s important to note that even the most heavily cited studies above had a sample size of less than 150 people. Furthermore, these studies interviewed participants directly about their adjustment to their disabilities, so they may have recruited a sample that was more outgoing or more invested in the research.
Another major issue with the above studies is that none of them had access to baseline (pre-disability) happiness. Other, more recent studies do, however. They found only a partial return to baseline happiness after disability. Richard Lucas examined the happiness and life satisfaction of people who became disabled, based on nationally representative studies of over 39,000 households in Germany and 27,000 households in Britain. From both data sources, Lucas found substantially lower levels of life satisfaction and higher levels of distress in people after they became disabled. He followed the German population for an average of 7.18 years before the onset of their disability and 7.39 years after, and the British population for an average of 3.48 years before the onset and 5.31 years after. People had a significant decline in happiness for the first year of their disability, and partially adapted, but did not return to their pre-disability baselines level of happiness. More severe disabilities were correlated with a lower final happiness level. These results held true even when controlling for employment and income changes.
Andrew Oswald and Nattavudh Powdthavee used the same British dataset as Lucas and found that those who became moderately disabled (able to still do most day-to-day activities, such as cooking, dressing oneself, and walking for ten minutes of less) return to about 50% of their previous level of happiness, and those who are severely disabled return to about 30%.
These studies have some notable strengths. They are based on large surveys of individuals not selected because of disability, and their level of happiness was followed before, during, and many years after developing their conditions. This allows for a more rigorous analysis of adaptation than matched control or retrospective studies.
It’s worth noting that the population in these studies that developed disabilities also had significant decreases in happiness as much as two years before their disability developed. This may reflect the unfortunate reality that people living in more difficult circumstances are disproportionately likely to develop a disability. So, it’s possible that some of the lower happiness measured after disability (relative to baseline) was not due directly to the disability itself, but to unobservable, progressive factors correlated with it.
Furthermore, disability is a broad term, and it is probable that those who have a disability due to a terminal illness experience greater decreases in happiness than those who develop a disability independent of another condition. The two studies above don’t differentiate between type or cause of disability.
Regardless, there is evidence that people psychologically adapt to their disabilities, but only partially. This isn’t to say that they are depressed – many still report extremely high levels of happiness (higher than many people who aren’t disabled, in some cases), just not necessarily as high as before developing their disability. Furthermore, it doesn’t mean people with disabilities lead less fulfilling lives. Like losing a loved one, developing a disability has an indelible effect on one’s day-to-day life. People with disabilities may not automatically return to their baseline happiness, but they can lead lives as fulfilling as those without disabilities, especially if they are given the support they need.
“I bet you $1,000 that if you just tell me you’re putting Demerol in my I.V. but don’t actually do it, I would still feel a lot better,” I dared the nurse.
It was a humid June night in 2013. I had just arrived at the emergency department of Mount Auburn Hospital in Cambridge, Mass., and was in the worst pain of my life. A kidney stone was scraping its way down my right ureter, the narrow tube connecting my right kidney and bladder.
Despite my anguish, I was so confident I didn’t need a narcotic painkiller that I was willing to put $1,000 on the line. The reason: I trusted the placebo effect, the relief of symptoms just by believing one is receiving helpful care. (In this case, even though I knew the nurse might be deceiving me, I was going to believe him if he said he was giving me Demerol. That’s essential for the placebo effect to work.)
My colleague Aaron Carroll wrote last week about the importance of placebo-controlled evaluation of medical treatments and devices. We need to know when medical care is better than a placebo.
In comparing a treatment with a placebo, we should also keep in mind that the placebo is not the same thing as the absence of treatment. In research settings, placebos are specifically designed to mimic treatment without the hypothesized few “active” ingredients or procedural steps. They still include a lot of components of care. (Another form of clinical trial is to compare one treatment with another or with “usual” care — the care that would be given in the absence of the treatment being tested.)
In evaluating surgical treatments, researchers go to considerable lengths to provide elaborate, sham procedures for comparison because care delivers cure by two pathways. One is through only our bodies — for instance, when surgery addresses the mechanical source of a physical problem. The other involves our minds: When we believe we are receiving helpful care, we get better. Moreover, we do so more quickly and at a higher rate than if we receive no care at all.
The question addressed by placebo-controlled trials is whether the second effect — the placebo effect that operates only through belief — is the only effect of a given treatment. Does the “active” part of the treatment do anything more? The possibility that placebos cure is therefore acknowledged and built into placebo-controlled study designs.
To be sure, some treatments are better than placebos. These cure through both pathways, including the one activated by placebos. At the same time, the placebo effect is not ubiquitous; some studies show no difference between placebo-controlled groups and no-treatment groups. Althoughsome of the ways in which placebos work are known — for instance by activating natural neurochemicals that make us feel good — we do not yet have a full explanation of when and how they do and don’t.
If placebos were always the same as no treatment, then the following findings, most of which are summarized by the emergency physician David Newman in his book “Hippocrates’ Shadow,” would be hard to explain:
- Taking two placebo pills (e.g., sugar pills) relieves more pain or provides a greater stimulative effect or is more sedating or heals stomach ulcers more quickly (depending on the study) than taking just one.
- Placebo pills with a brand name printed on them are more effective at pain reduction than the same pills without the brand name.
- Patients who faithfully take placebo medication for cholesterol reduction survive longer than those who skip doses.
- Though sham acupuncture reduces migraines as much as real acupuncture, both reduce migraines far more than no treatment at all.
- Measurements of increased endorphins — our bodies’ natural pain relievers — have been associated with placebos’ ability to reduce pain.
The hypothesis that when we believe placebos will heal, they do, at least to some extent, is hard to reject. Perhaps for this reason, our childhoods are full of placebo effects, and as parents, we deliver them to our children. After a boy’s minor fall, which mainly hurts his pride, a bag of ice on his knee soothes even if the knee really isn’t injured. A bandage over a girl’s scrape that didn’t break the skin ends her tears. Hugs and sympathetic tones go a long way. My mother once cured one of my childhood headaches with a piece of cheese that she said some people thought effective at doing so. She made that up, but I believed her at the time. These are all placebos.
Given the strength and ubiquity of placebo effects, many physicians prescribe them. In fact, doing so was common practice before World War II, with supportive publications in the medical literature as late as the mid-1950s. This practice faded away after the rise of placebo-controlled trials that yielded treatments that were shown to be better than placebos, but it has resurfaced in new forms.
Today, the widespread use of antibiotics for conditions that don’t require them is a form of placebo prescribing, for example. Acetaminophen for back pain appears to be a placebo as well. These may help patients feel better, but only because they believe they will do so. The active ingredient adds nothing. To the extent some doctors trick patients in an effort to achieve a placebo effect, most patients don’t seem to mind. Nevertheless, deliberately harnessing just the placebo effect by prescribing a treatment that does not have any additional direct physical effect is an ethical gray area.
The lesson of placebos is simple: The mind-body connection is strong. A lot of good can come from caring and feeling cared for. Sometimes we need, and can find, additional help from surgery, medication and other therapies. But for a wide range of common problems — from earaches to knee pain to headaches — sometimes we don’t.
When a clinical trial tells us that a therapy is equivalent to placebo, that doesn’t necessarily tell us that the therapy does little; it may tell us that the placebo does a lot. The therapy merely does nothing additional.
“I’m giving you Demerol,” the nurse said. (He didn’t take the bet.)
Within seconds, the pain from my kidney stone subsided. But, in fact, I had already begun to feel significant relief from the intravenous fluids alone — just saline. My pain, I surmised, had been exacerbated by the panic of being vulnerable and at some remove from care — first at home, where the pain started, and then in the car on the way to the hospital. Just believing I was safe and in a place where any problems could be addressed went a long way.
Though the Demerol is stronger than the placebo effect alone, I may not have needed it to reduce my pain to a tolerable level. But I definitely needed and received care, and felt some relief from that alone. The placebo effect did a lot of the work.
My health law students and I were discussing HIPAA’s Privacy Rule when we got to talking about the iCloud hack of the nude celebrity photos. Although publication of the photos was a grotesque invasion of the celebrities’ privacy, there’s been no big push for the federal government to pass a law requiring Apple to take better care of its customers’ data. Yet the federal government already has a law in place—the Privacy Rule—that closely regulates how health-care providers protect their patients’ medical records.
Why do we treat medical data so differently from other personal data? The answer isn’t obvious, at least not to me. It’s probably not because medical data is categorically more sensitive. When I asked my students whether it’d be a bigger invasion of their privacy to search their medical records or whatever data their phones sent to the cloud, their answer was unequivocal and universal: the cloud data.
I think that’s what most of us would say, at least those of us with one foot in the digital age. That’s not to minimize the importance of medical privacy. But just think about the snarky emails, off-color text messages, and (maybe) naked selfies you’ve got floating around in the cloud. Are those really less sensitive than an embarrassing medical condition? For some people, sure. But for most?
Maybe there’s a better reason. Studies consistently show that people are really, really worried about medical privacy. Without HIPAA’s Privacy Rule, maybe people would avoid doctors or refuse to tell them everything they need to know. That’d be a problem. In contrast, people who fear the disclosure of their other personal data can always unplug from the cloud or change their privacy settings. No big deal.
There’s something to this, but I’m not sure it stands up to close analysis. For one thing, I doubt that members of the public know enough about the Privacy Rule to change their behavior. For another thing, participating in digital life isn’t really a choice. It’s a fact. It’s perfectly reasonable to think that the price for connecting to the cloud shouldn’t be the surrender of personal privacy. If we need a law to get people to go to the doctor, why wouldn’t we also need a law to make people comfortable with the cloud?
I can think of one other reason for treating medical information differently. Technology companies like Google, Apple, and Facebook know that they’ll lose customers if they’re careless with personal data. Even when data breaches happen, the story goes, we can trust they’re doing all they can to stop them. Maybe medical providers don’t face the same incentives. They’ll be reckless with medical data because the market won’t adequately punish them if they’re not. Like sheep, the patients will still come.
But is that right? It’s certainly plausible. Still, I wonder. It’s not like providers don’t care about the market. Hospitals cultivate their reputations, for example, because those reputations affect their bottom lines. They can’t just brush off newspaper exposés about data breaches. When Stanford Hospital in Palo Alto accidentally posted data on 20,000 ER patients, or when a laptop containing 33,000 medical records was stolen from a Cedars-Sinai employee, the hospitals didn’t just worry about the potential HIPAA penalty. They also worried that the scandals would drive away patients.
Sure, the incentives for technology companies might be sharper than for health-care providers. Patients can’t pick where they get treatment as easily as they can pick which company carries their personal data. Yet, as Richard Epstein has observed, there was no “explosion of improper disclosures of sensitive [medical] information” prior to the advent of the Privacy Rule, in part because providers were already subject to state privacy laws. And I’m not aware of evidence that hospitals and doctors were systematically cavalier about patient records. (Lurid anecdotes don’t count.) In any event, it’s not like Facebook, Apple, or Google have always put their customers’ privacy first.
So I guess I’m left scratching my head. If we think that HIPAA’s Privacy Rule makes for good policy, shouldn’t we also consider a similar rule for companies that have custody over cloud data? Alternatively, if we think it’d be counter-productive to superintend data privacy at Apple, might it also be counter-productive to do it at hospitals and physician practices?
The US has reduced its commitment to health research. The decline in support for health research has many consequences, none of them good. One consequence may be to concentrate health research funding in a small number of institutions.
You can see the overall decline in this graph (from Harvey Fineberg):
The next graphs describe the trend at Johns Hopkins University (from grantome.com).
The upper left graph presents counts of R01 grants — the principal NIH research grant — given out since 1985. The upper right hand graph shows the trend for Hopkins. The absolute number of R01s going to JHU has recently declined, mirroring the overall trend.
Now Hopkins is a special case: it receives more R01s than any other university or hospital (2.4% of all R01s). The graph on the bottom shows that despite the absolute decline, JHU’s R01 share has been steady for a long time. Grant competitions are zero-sum games and Hopkins is holding its own against the competition.
There is an interesting trend in the relative positions of the top 100 universities dominating biomedical research.
The vertical axis is the proportion of the total R01s received by an institution in 2013. The horizontal axis is the rate of change in that proportion, with institutions that have increased their share of funding to the right. The vertical axis is the change from 2010 to 2013 in the proportion of graphs held by the institution. The folks at grantome.com interpret this as showing that the rich are getting richer.
You have probably noticed a problem with this graph. To show that the rich are growing richer, grantome.com ought to have looked at whether the proportion of grants held in 2010, not 2013, was associated with the change from 2010 to 2013. But let’s suppose that the corrected graph would show something like the above pattern: that the institutions that already have the most grants are increasing their shares of the R01 funding.
Why would that happen? Researchers get a lot of advantages from being at a research-focused school or hospital. Biomedical research is multidisciplinary and at a big shop it’s easier to find highly talented specialists with the skills you happen to need on your research team. Big research factories may provide better mentorship for junior researchers. Conversely, many of these schools ruthlessly cull faculty who lose grant competitions. As funds dwindle and the competition gets more intense, perhaps these advantages become more important.
Is the concentration of the grant wealth in fewer hands good or bad for science? I have no idea. Maybe the most efficient way to carry out biomedical research is to have a relatively few large super-universities
But if you look at the graph of the 50 most successful universities, something else leaps out. 42 of the top 50 are in ‘blue states’ (that is, carried by Obama in 2012), including the top 13. Most of these top schools have seen an increase in their R01 share since 2010. Five of the 8 red state schools have seen a decline in the proportion of R01s they received.
This isn’t because Obama has steered funding towards ‘his’ states. I’ve served on many NIH grant review committees. Politics never comes up.
More importantly, the pattern can’t have much to do with contemporary politics. The advantage held by northeastern and west coast schools has been around forever. It stems from — but also partially explains — the historical economic advantages of the blue states. You can see the geographical disparities in scientific achievement in other data. Richard Florida mapped the per capita distribution of scientific citations.
The colored peaks are concentrations of scientific productivity. Thomas Friedman had claimed that the world is flat: you can live anywhere and be part of modern scientific culture. Florida showed that this is absurd: the scientific world is spiky. Post-war global science has been led by the US, but not really. It’s been dominated by Massachusetts, the Washington-NYC corridor, and California.
Here are my concerns about the rich getting richer. First, increasing concentration of science on the coasts will increase US regional economic and educational disparities. Red state members of Congress cutting the NIH are hurting their constituents’ children. Second, a greater concentration of scientific dominance in a few liberal states is to the disadvantage of the NIH, because over time it must further erode broad political support for medical science. And what diminishes the NIH is greatly to the detriment of the nation as a whole.
UPDATE: Edited cause I totally misread an indicator!
This time it’s obstetric quality metrics. “Association Between Hospital-Level Obstetric Quality Indicators and Maternal and Neonatal Morbidity“:
Importance In an effort to improve the quality of care, several obstetric-specific quality measures are now monitored and publicly reported. The extent to which these measures are associated with maternal and neonatal morbidity is not known.
Objective To examine whether 2 Joint Commission obstetric quality indicators are associated with maternal and neonatal morbidity.
Design, Setting, and Participants Population-based observational study using linked New York City discharge and birth certificate data sets from 2010. All delivery hospitalizations were identified and 2 perinatal quality measures were calculated (elective, nonmedically indicated deliveries at 37 or more weeks of gestation and before 39 weeks of gestation; cesarean delivery performed in low-risk mothers). Published algorithms were used to identify severe maternal morbidity (delivery associated with a life-threatening complication or performance of a lifesaving procedure) and morbidity in term newborns without anomalies (births associated with complications such as birth trauma, hypoxia, and prolonged length of stay). Mixed-effects logistic regression models were used to examine the association between maternal morbidity, neonatal morbidity, and hospital-level quality measures while risk-adjusting for patient sociodemographic and clinical characteristics.
Main Outcomes and Measures Individual- and hospital-level maternal and neonatal morbidity.
These researchers picked two perinatal quality measures. One was the percentage of deliveries that were elective before 39 weeks gestation. The other was c-sections for low-risk mothers. You want both to be low. Then they looked at both maternal and neonatal morbidity, and measured whether they were correlated with “quality”.
Here’s neonatal morbidity:
Quality is on the x-axis. Mortality is on the y-axis. Can you see a relationship? Cause I can’t. Neither could the statistics. The quality measures “were not associated with severe maternal complications (risk ratio [RR], 1.00 [95% CI, 0.98-1.02] and RR, 0.99 [95% CI, 0.96-1.01], respectively) or neonatal morbidity (RR, 0.99 [95% CI, 0.97-1.01] and RR, 1.01 [95% CI, 0.99-1.03], respectively).”
I don’t dislike the theory behind “pay for performance”. I just don’t think that the things we measure are often really “quality”. And the literature keeps making me feel like I’m right.
Below are my notes from reading David Cutler’s The Quality Cure. Indented bits are paraphrases. Block quotes are direct quotes, obviously. Unintended bits are my commentary. Hyperlinks added to quotes are mine, based on references in the text. Asterisks indicate areas I intend to look into more fully in the future.
If you want my short take, skip to the final paragraph of this post.
Unnecessary care or overuse of certain therapies adds up to nearly $200 billion, according to Berwick and Hackbarth. Treatment of back pain is a classic example. Spine surgeries are often not necessary but are done in the U.S. at twice the rate as they are in other countries. There is a sixfold difference in back surgery across U.S. regions, only 1/10th of which is due to patient factors.
Induction of preterm birth is another example. Elective induction of birth should only be done for after 39 weeks of gestation, but 40% are earlier than this.
A third example is stenting.
The point is that care is appropriate in some cases but inappropriate in others, and the medical system does a poor job of separating appropriate use from inappropriate use. To use an analogy, health care waste is like fat layered into beef. One cannot remove it by simply cutting entire slabs. Rather, a delicate and deft knife is needed to separate the good from the bad.
It’s examples like these that cause the number needed to treat (NNT) to achieve one good outcome (relative to the counterfactual of no treatment) to be higher in practice than in clinical trails in which therapies are more targeted to appropriate populations. About elective induction of pre-term birth, see Aaron’s recent post.
There are protocols that medical personnel can follow to essentially eliminate [central line] infections. Peter Pronovost of Johns Hopkins University has pioneered the use of these protocols and has demonstrated that institutions can use them to essentially eliminate infections.
The fact that this is possible is consistent with the idea that health care organizations vary in productivity (quality), or that spending and health outcomes are meaningfully related to provider factors. If spending and outcomes were nearly entirely driven by patient factors, then protocols like this wouldn’t be of much use. One can make the same point with another example found later in the book: implementation of a Crew Resource Management (CRM) program in labor and delivery at the Beth Israel Deaconess Medical Center in Boston that dramatically reduced adverse obstetric events and malpractice claims (pages 132-134).
[T]he death rate from medical errors [...] is the equivalent of a medium-sized jumbo jet crashing daily.
[O]bjections notwithstanding, I have added Canada and East Asia to “Europe.”
For humor value, this may be my favorite line of the entire book.
All high-income people in the United States earn more than high-income people in other countries, and thus the pay of physicians is not out of line relative to other highly educated people. [...] Clearly one reason for high medical spending in the United States is that income distribution as a whole is more unequal in the United States, and health care uses many highly skilled workers. There is little the health care system can do about this—though overall income distribution is certainly responsive to policy.
I read this as the opportunity cost point that John Goodman is fond of making. But this does not imply that we cannot have better care that is also cheaper per (good) outcome. Even at current prices and wages, there is a lot of overuse, underuse, and misuse.
From the link, it appears as if the U.S. is actually in third place, behind Germany and Belgium, in stent insertion rate in 2009. By 2011, the U.S. rate had come down relative to OECD countries and was in 7th place. The U.K. inserts stents at nearly half the rate as the U.S., just like Canada. Mortality after a heart attack is considerably higher in the U.K. than the U.S., however. In Canada, it’s about the same. Not controlling for anything, it’s hard to draw any conclusions about the right rate of stenting. More international comparisons of quality for patients with chronic conditions here.
The canonical medical treatment in the direct-marketeer framework is Lasik surgery [for which most] people pay out of pocket. Over time, the price of Lasik has fallen, even as quality has improved. Wouldn’t that be true of all medical care if consumers were in charge? [...]
Move away from Lasik, and the world suddenly becomes less clear. Dental care is not well covered by insurance, and the environment is leading to healthier teeth, yet the costs of dental care have increased nearly as rapidly as the costs of medical care. Even veterinary costs are increasing over time, at roughly the rate of human care, and very few people have insurance for their pets.
This actually makes the point that opportunity cost isn’t necessarily the only relevant consideration for price reductions and quality improvement. It happened in Lasik! But, Cutler’s point is that that fact does not imply that remaking all of health care to look like the Lasik market would work.
Put simply, no industry ever got better without knowing what it was doing.
This is in the context of low use of information technology in health care.
Cesarean sections are reimbursed more generously than vaginal births, however, so cutting back on [unnecessary] cesarean sections reduced revenues to Intermountain. Insurers saved money, but Intermountain did not. The same was true at Virginia Mason when they reduced the number of MRIs and orthopedic consults for lower back pain. [...]
The net effect is that most clinical savings gained by operating hospitals more efficiently are not realized by the hospital undertaking the investments. As a result, hospitals tend to ignore efficiency.
This is in the context of why hospitals might not invest in IT (EMRs) on their own. Contrast with the recent findings of Dranove et al., however.
About a RAND study that concluded in 2012 that investment in health IT had been “disappointing” with only “mixed” impact on efficiency and quality, Cutler wrote,
As of 2012, when the assessment was made, health IT was still in the dissemination phase. The technology was nowhere near ubiquitous [...] systems were not [...] interoperable, [...] appropriate workflow changes [had not yet been made]. It takes a while to learn how to use an entirely new system and reconfigure practices to benefit from it. I expect that seven years from now , the conclusion about health IT will be very different.
There is no perfect system lurking out of sight.
Pages 117-119, 126:*
The growth and success of bundled payments.
It’s not yet clear from Medicare demonstrations whether “lower-performing organizations can transform themselves into higher-performance ones as the payment model changes.” Savings from the physician group practice demonstration was meager, for example.
However, Blue Cross Blue Shield of Massachusetts’ Alternative Quality Contract has achieved more impressive performance. Reasons could include that private prices vary more than Medicare prices, so switching to lower cost providers is more feasible; AQC participants were less integrated, providing more opportunity for care management and coordination; The AQC rolled out during a time of intense focus on costs in Massachusetts.
States are also pushing new payment models.
Cutler explicitly mentions Arkansas, but Illinois, Oregon and other states also come to mind. Note also that alternative payment models promoted by one entity (like Blue Cross Blue Shield) can have spillover effects, altering the cost and quality of care for individuals covered by a different entity (like Medicare). On this point, see the work of McWilliams, Landon, and Chernew.
Payment reform may thus become the province of state policy.
One way or another, payment systems for medical care are likely to change markedly in the next few years. [...] My guess is that payment will change substantially within the next five years [by 2019], and the fee-for-service system will be largely gone within a decade [by 2024].
Two things could go wrong: (1) Insurers could work at cross-purposes, each paying in different ways that conflict. (2) Larger hospital organizations are in the best position to organize around new payment models but those models also tend to make hospitals the least efficient setting for care delivery. Physician-led organizations may make more sense, but it is unclear how they will emerge.
The global payment model is fundamentally different from the HMO model. In an HMO, insurers dictate to doctors and patients what they are allowed to do and what they cannot. [...]
In the global payment model, in contrast, physicians decide on good care and work with patients to provide that care.
Phase 1 [of health care reform] is covering people, and phase 2 is reforming the payment model to encourage better care. Phase 3 involves organizations changing their internal structures to deliver higher-quality, lower-cost care. [...]
The good news is that the difference between good and bad performers is not a wholly different set of employees. [...] Rather, the differences have to do with how the organization defines its mission, measures what it does, trains its employees, and motivates them.
An instrument developed by McKinsey asks organizations about performance monitoring, target setting, and incentives/people management. In high performing organizations, information relevant to performance is fed back to workers who are empowered to stop processes to fix problems; goals are established to focus attention on areas for improvement; people are hired and promoted on the basis of performance.
Bloom et al. used the McKinsey survey instrument to examine over 10,000 organizations internationally, mostly manufacturing firms, but also several hundred hospitals and some schools, including those in the U.S., U.K., Japan, Germany, and some developing countries.
In manufacturing, firms that score better are also more profitable and successful. Higher scoring schools do better on standardized tests. Higher scoring hospitals also have better survival rates for heart attacks. Across all three domains, U.S. hospitals have lower management scores than manufacturing firms, but higher than hospitals in other countries. More details here and here and here.
[T]he most important rule of health care management is this: never put care providers in a position of denying care for financial reasons.
In contrast to the U.K., Cutler does not advocate denying coverage for high-cost, low- (but not zero) value therapies (like Avastin for colorectal cancer, which can extend life by 1.4 months for $50,000) at this time. He recommends we focus on reducing pure waste (zero value care) first.
At some later point, we might need to consider whether Avastin is worth covering as a central benefit, but deferring this decision is reasonable.
The second rule of health care is:
[I]ndividual physicians should not be compensated based on the clinical outcomes of each patient.
Instead, Cutler recommends aggregating over physician groups and the population of patients they serve.
The third rule of health care is:
Patients have a lot to contribute to care improvement, and their voices should be heard.
About medical malpractice reform:
I fully admit that we do not have the answers yet.
About creating a more efficient health care system:
Unfortunately, there is little that society can do to force this change. What we can do is set the stage for it.
Focusing on quality has the potential to fix much of what ails American medicine. But how long will the quality cure take? [...]
No one knows.
This is how health care will become more productive and Cutler’s guess about how long it will take:
Economists Stephen Oliner, Daneil Sichel, and Kevin Stiroh studied the sources of the increase in productivity growth over the past two decades. [...] They found that higher IT use was associated with more rapid growth in productivity. Industries that used IT above the median level grew 1.5 to 2.0 percentage points more rapidly than industries that were low users of information technology. Here, then, is our first metric: as we move health care from an economic laggard to a leading industry, [productivity] growth might increase by 1.5 to 2.0 percentage points annually.
There is no citation for the Oliner/Sichel/Stiroh paper, but here’s one by them that seems relevant to the discussion. (I have not read it.)
Another way to gauge the potential in health care is to look at what needs to be done and estimate how rapidly it can occur. [...] The easiest changes are in the site of care. This involves people who are being hospitalized in expensive institutions when they could be treated just as well in less expensive ones or even on an outpatient basis. [...] The groundwork to affect [this] could be laid within one to two years [by 2016]. [...]
Somewhat more difficult are changes that need to occur within institutions, to streamline the pathway of care for patients with various conditions [...] rationalizing who receives stents and who does not, implementing care pathways for routine labor and delivery, [etc.]. [...] My guess is that three to five years of work are required before major savings from these pathways can be realized [by 2021 if these follow after site-of-care changes].
The third tier of savings comes form populationwide prevention and patient engagement. [...] Such experimentation will need at least five years to start bearing fruit and likely a decade before major savings can be realized [by 2031 if this follows prior changes]. [...]
All told, therefore, improving health care quality is a fifteen- to twenty-year venture. If we are able to pull out 30 percent of costs in fifteen years, this implies a cost reduction [productivity increase] of 2 percent annually. If the transition takes 20 years, the implication is an average cost savings [productivity increase] of 1.5 percent annually. Here is the second piece of evidence: health care reform will increase productivity by perhaps 1.5 to 2 percent annually for fifteen to twenty years.
If this productivity growth were entirely achieved by (or translated to) reductions in spending at the same rate, this would probably bring overall health care spending in line with GDP growth. However, as Cutler points out, we see higher productivity associated with more overall spending in other industries. Michael Chernew made this point about IT: we spend a lot more on it than we used to even as each IT-related product or function becomes cheaper, controlling for quality. (Today’s cell phones do a lot more for their price than do those of a decade ago. But we may spend more overall on cell phones, in part for that reason.)
No one knows for sure how near-term changes will play out, let alone their longer-term effects. But in the end, there is certainly reason for optimism. If we do things right, the future of health care could be very bright indeed.
I thought one of the best aspects of the book was the expression of optimism and realism throughout—evidence-based and without overbearing cheer leading. Too many health policy books take a grumpy “it’s all terrible” tone. Too many also suggest solutions that are politically unrealistic. Cutler’s is decidedly different. He’s neither grumpy nor naive about what’s possible. I also liked that the book didn’t belabor any points. At 171 pages (of main text), given what it covers, it is laudably efficient. Few books are.
* More about these areas another time … maybe.