Is More Medical Testing Better?

“I think this patient needs a CT scan of her chest,” the consulting
physician said to me over the phone. “Her lungs sound bad, and given her
history, we need to make sure she isn’t developing pneumonia.”

The patient, though only thirty-five, had been through a lot during her
hospitalization. She came in a week prior with some shortness of breath, an
abnormally high heart rate (tachycardia) and chest pain. But her heart
checked out just fine. An EKG was normal and her troponin, a protein
that spills into the blood with damage to heart, was undetectable. Because
of her tachycardia and chest pain we tested her for a

pulmonary embolus
, or clot in the lungs, with a CT scan. It was positive. The potentially
deadly clot blocks off blood flow coming from the right side of the heart
into the lungs causing heart strain and disrupting oxygenation of blood and
the functioning of the heart.

We admitted the patient to the

intensive care unit

where she received IV medication to thin her blood. She recovered but
continued to experience residual chest pain that would clear in the months
ahead. As we prepared to discharge the patient, the consulting physician, a
cardiologist, told us he wanted her to get another CT scan of her chest.
She had not had a scan since the initial stages of her workup. What if she
developed pneumonia in the interim? Her white blood cell count, often a
crude marker of infection, was not elevated. She did not have fevers. When
asked, she felt well enough to go home and wanted to leave. Then again,
being in the hospital made her susceptible to infection. Moreover, on her
physical exam, we heard crackles in her lungs –
this sometimes indicates an intrapulmonary pathology. The cardiologist’s
concern gave us pause and we ordered the imaging study.

~

Doctors often feel uncomfortable with areas outside of our expertise. Consequently, we call other specialists to see the patient and give us advice. Moreover,
patients sometimes ask to see a specialist in the hospital: “Can you call
the neurologist to come see me?” or “We’d like you to call a cardiologist
to see our father while he’s here.” Because consultants share a different
knowledge base than the team primarily caring for the patient they may ask
for more tests to rule out other serious pathologies that the primary team
neglected to consider.

The patient’s repeat CT scan merely demonstrated small collapsed alveoli.
These terminal branches of the lungs often collapse when we draw shallow
breaths or lie flat for a long time, a typical finding in many hospitalized
patients. Given the benign nature of this finding, we discharged the
patient. Yet she had received an extraneous dose of radiation and her
hospital bill would be hundreds of dollars more. Did she absolutely need
this? This common story raises other questions, too. Do patients do better
with more specialists seeing them? Do patients do better with more testing?

In a 2012 post for the New York Times Well blog, Tara Parker-Pope

pointed out

that “overtreatment – too many scans, too many blood tests, too many
procedures – is costing the nation’s healthcare system at least $210
billion a year, according to the Institute of Medicine.” And the stories
she tells about astronomical hospital bills due to overtesting are
disturbing.

In
a 2015 article in the Journal of the American Medical Association, a group of researchers found mortality for high-risk
heart failure and cardiac arrest among patients was lower in teaching hospitals
during national cardiology meetings compared to the rest of the year – meaning that the absence of a large number of cardiologists, who were attending meetings, was correlated with lower mortality for these heart conditions in the hospital. In an editorial in the same issue, Dr.
Rita Redberg makes a disquieting suggestion: “How should we interpret these findings? One possibility is that more interventions in high-risk patients with heart failure and cardiac arrest leads to higher mortality.” Is there too much being done, especially by experienced
physicians?

Dr. Ezekiel Emanuel, an oncologist and bioethicist, elaborated on these questions in a New York Times op-ed in 2015:

We – both physicians and patients – usually think more treatment means
better treatment. We often forget that every test and treatment can go
wrong, produce side effects or lead to additional interventions that
themselves can go wrong. We have learned this lesson with treatments like
antibiotics for simple medical problems from sore throats to ear
infections. Despite often repeating the mantra “First, do no harm,” doctors
have difficulty with doing less – even nothing. We find it hard to refrain
from trying another drug, blood test, imaging study or surgery.

When specialists like neurologists or cardiologists see a patient, they
approach the bedside from a unique perspective. The pathologies they know
and think about are very different from what family medicine or internal medicine doctors thinks of when they see a patient. Specialists, who often act as consultants, consider the diseases they are most worried about within their field. They’ve been asked to see the patient to
recommend workup for a disease potentially related to
their area of expertise. Their view, in other words, is necessarily myopic – if you give a carpenter a hammer, surely the carpenter will find a nail.
This does not always happen, but by nature there is a bias when a consultant
approaches a patient – and that bias is toward ordering another test,
toward doing something. Part of the art of medicine, especially as a specialist or consultant, is figuring out when the patient needs
something and when the best approach is to do nothing at all. Our patient
at the beginning of the story did not really need a repeat CT scan. To be
sure, the cardiologist didn’t recommend it simply to radiate the patient or
increase the hospital bill. But none of us wanted to miss something.

A conservative method of practice can come with experience, but as evident
from the JAMA study referenced above, that’s not necessarily the
whole story. Some of this, I think, requires thinking regularly about how well the
patient in front of the doctor is doing and how a test will change the
course of the patient’s treatment. “Will this change our management?” is a
question our attending physicians always ask us before we order a test. And it is a
question all doctors must ask themselves.

When Doctors Are Wrong

As medical students and resident physicians gain experience they also gain
knowledge and confidence. Consequently, young trainees eventually reach a
level of comfort in speaking with families and patients about prognosis and
disease course. This is part of the purpose of training, as these
conversations happen so often that they are an integral part of practicing
medicine. But it isn’t certain that with experience and confidence also comes
accuracy.

***

Soon after finishing sign-out on a night shift I received a page. The
nurses told me that a patient’s daughter had arrived and wanted to speak
with me about her father. The patient was an elderly but relatively healthy
gentleman who had been admitted with abdominal pain. Multiple imaging
studies had shown little to account for his discomfort. But the pain was so
intense that he could not eat anything. Even going to the bathroom was
difficult — he held in his stool to avoid the agonizing act of defecating.
As a result, he became constipated, which then exacerbated the pain. Thus,
the medical team administered anti-inflammatory treatments, stool softeners, and IV
fluids as they searched for the etiology of this troublesome symptom.

The patient’s daughter asked me about the latest imaging studies and labs
as she sat holding her father’s hand. Buried to his chin under the covers,
the patient participated in the discussion, asking when he would be able to
eat and go to the bathroom easily again. I explained that all the tests had
been negative so far and we were unsure of what was going on. I then left the
room, and the daughter caught up to me in the hallway. She seemed worried,
speaking hurriedly and pleadingly: “How long do you think my father has
left to live? How much time do I have left with him? Should I start making
funeral arrangements?”

The questions took me aback. I had seen plenty of patients in the
intensive care unit
who were on the verge of death and they looked so different from her father
— a profound pallor, somnolence, lethargy, disinterest in conversation
and food. But this patient wanted to eat again, he wanted to see his
family, he wanted to watch basketball on TV, and he was interactive and
conversational. His cheeks certainly did not exhibit the deathly pale hue
of those crossing over to the other side. I assured the daughter of this:
“I don’t think you have to worry about that. The most important thing right
now is that we figure out what is going on. I can’t tell you how long he’s
going to live but I would be shocked if he had only days or weeks left.”

This interaction repeated itself for the next three nights, always with the
same diagnostic uncertainty. On further imaging there was evidence of some
abnormal fluid in the abdominal cavity. Interventional radiologists
extracted the fluid to test it for any cancerous or infectious cells, but it
would take perhaps a week or more for definitive results to come back. In
the meantime, the patient’s pain improved and he moved his bowels without
issue. Even though we didn’t yet have a diagnosis he seemed to be doing
incrementally better each day.

On the fourth night I again saw the daughter and she asked: “You don’t
think I should be planning the funeral for my father, do you? He’s not
going to pass in the next few days?” I understood why she was asking the
question — any child ought to be concerned for the well-being of a parent.
But I was also surprised because her father was on the mend. I told her
that if he continued to improve he would leave the hospital, and his primary
care doctor would follow up the lab results and see him in clinic.

At around 2 a.m. that same night, a voice over the hospital loudspeaker
echoed throughout the halls as I jumped out of my chair: “Code blue, 7th floor, code blue, 7th floor.” There was a
patient in cardiac arrest. I ran out of the workroom and met up with
another resident. Which patient was dying? On my way to the code I ran
through the patients on the coverage list that evening; I did not expect
anyone to pass away. As the other resident and I ran down the hall I saw
the code cart containing all the medical resuscitation equipment necessary
to treat cardiac arrest outside of the room I had visited every night for
the past four nights. My heart leapt out of my chest; I pleaded with some
higher power that it not be that patient. But it was.

The resident, nurses, and I immediately began CPR. The anesthesiologists
burst into the room and stuck a tube down the patient’s throat and into his
trachea to protect his airway as the respiratory therapist attached the
tube to a ventilator to help the patient breathe on his own. After multiple
rounds of CPR, his pulse returned. We wheeled the patient — attached to
tubes, and poles filled with intravenous fluids — to the ICU for closer
monitoring. He didn’t respond to our questions or poking and prodding, but
he was alive.

Alas, as soon as we got to the ICU, his heart once again
stopped beating and his IV line ceased working — his veins (which can
happen as we age) were friable and brittle, and the small vein carrying the
volume and force of the IV infusions burst. Without an IV we could not give
medications. We turned, then, to an

intraosseous (IO) line
. This entails drilling a hole into the bone and infusing medications
through that hole. It is a proven method of administering medications when
physicians and nurses cannot obtain IV access. I opened the IO kit and
attached the drill to the IO needle, placing it on the shinbone and drilling. It
slid into the bone, I detached the drill, and hooked up the
IV tubing to the IO line jutting out of the patient’s bone. At this point,
the code had been going on for nearly 15 minutes and the patient’s family
had arrived. They watched as we furiously attempted to revive their loved
one. At some point a family member shouted “Stop, please, enough!” Time of
death: 2:45 a.m.

I sheepishly held my head down, avoiding eye contact with the family as
they sobbed. The medical team and nurses quietly left the room, leaving the
patient in peace. As I passed by the daughter, I could only say “I’m so
sorry” — little else would have sufficed. Not only did we not save him, but
night in and night out I had given the daughter a false impression that he
wouldn’t die. Perhaps, I wondered, I had been disingenuous in some way.
Either way, I was wrong.

 ***

Alas, physicians are wrong relatively often, and there is ample
evidence for this. In a systematic review in the

British Medical Journal in 2012
, researchers found that each year up to 40,500 adult patients in American ICUs die with a misdiagnosis. The Journal of the American Medical Association published an analysis in 2009, concluding, among other things, that “while the exact prevalence of
diagnostic error remains unknown, data from autopsy series spanning several
decades conservatively and consistently reveal error rates of 10% to 15%.”
The American Journal of Medicine published a

separate analytic review article in 2008
, concluding that diagnostic error occurs up to 15% of the time in most areas
of medicine. The authors further theorized that overconfidence often
accounts for at least some of the errors. These reports have reached a wide
audience in the laymen’s press as well. In 2015, the Washington Post published an article indicating that diagnostic errors affect 12 million adults each year. The
impacts of errors, as we see in the story above, don’t just involve the patient
but the patient’s families, too.

Though these statistics are shocking, it is almost impossible, from the
patient perspective, to look at them and subsequently be skeptical of everything a doctor says. After all, we are not only practically but also emotionally dependent on them: We want reassurance from our
physicians and we want definitive answers. As a patient, it is frustrating
to hear “It may or may not be cancer and we can’t be sure” or “I don’t know
how much longer she has left.” Indeed, when the path ahead of us is no
longer clear, we turn to physicians for answers because of their
experience. We want them to be the kinds of people none of us can
be — always right, always knowledgeable, always calm and composed. But they
are fallible, despite the impossibly difficult and long road they’ve
traversed.

And what can we as physicians take away from this? Doctors want to be
the kinds of people their patients expect them to be. But the statistics
of medical errors are the reminders of how impossible that is; how many years of studying and
experience are necessary even in order to be competent; how difficult,
despite the many exams we take and pass, it is to apply knowledge
appropriately. Not only are we fallible, but the science we rely on is not
always helpful either. Indeed, the best studies are useful merely for inferring what will
probably happen — they do not tell us definitively what will happen to the
patient in front of us. Moreover,

scientific evidence does not exist
for every treatment in every situation or every diagnosis in every
situation. Once again in medicine,

our ideal does not match with the real
, and our preconceived notions are sometimes shattered in moments of frustration
and uncertainty. Perfection is unattainable, but we must
constantly seek it out, always aware of how out-of-reach it lies.

When patients and their families now ask me questions about prognosis or
treatment I always preface what I say with: “Nothing is 100% in medicine.”
Though I will be wrong again in my career and will, hopefully, learn from
my mistakes, I never want to give a false impression. We often tend to ignore uncertainty or wish it away, but we must always remind
ourselves, whether as patients or doctors, that no doctor and no science is perfect.

The Problem with the New Patient Autonomy

The neurology team shuffled single-file into the patient’s small room. The patient, probably in his 30s, had black hair, brown eyes, and an unsettling
demeanor. He glared icily at us from his bed, the blankets covering him up to the neck. His pale brow furrowed even more noticeably as all nine of us
intruded on his privacy. In a scene out of a futuristic movie, EEG (electroencephalogram) leads on his scalp connected his
head via wires to a screen showing squiggly lines representing brain activity; a small video camera attached to the screen monitored the patient’s movement. He had come to the hospital overnight after falling and shaking, a story worryingly suggestive of a seizure.

Brain waves on EEG
Image via Shutterstock

An electroencephalogram records neuronal signals in the brain and is used by neurologists to diagnose seizure activity. When a
patient has a seizure, which can manifest as full-body convulsions, a family member in the room pushes a button on the machine which starts the video
camera recording the patient’s movements. Then, neurologists examine the movements in the video and the waves tracked by the EEG to see if they are consistent with seizures.

There are different kinds of seizures depending on which part of the brain is affected. Symptoms range from a loss of attention for a few seconds (absence seizures) to full-body convulsions
which we typically associate with seizures (generalized tonic-clonic seizures). Different conditions can cause these events — for instance, high fever as a child (febrile seizures) and brain tumors can induce
hyper-excitability in the brain. If the seizure does not stop, a patient can enter status epilepticus, a state of prolonged epileptic activity that can cause permanent damage.

Having a seizure, then, can be very serious business. Physicians must perform a medical work-up to ensure that the patient is not at great risk. In
addition to an EEG, our patient’s neurologist ordered labs and a CT scan of the brain. However, these tests were all negative. Even overnight, when the patient and his mother both claimed that the patient seized, there were no abnormal
electrical discharges on the EEG.

Indeed, not all physical manifestations of seizures indicate the presence of legitimate seizure activity in the brain, which is why the EEG is such a valuable diagnostic
tool. It turns out that certain patients may believe they are having seizures when they are actually having pseudoseizures or psychogenic non-epileptic seizures. To most observers,
pseudoseizures look exactly like generalized tonic-clonic seizures. Patients shake, tense up, and flail violently and frighteningly. However, certain
differences exist that distinguish them from each other. During pseudoseizures, EEGs show no abnormal brain activity, patients do not bite their tongues (this can occur with real seizures), and patients do not
respond to anti-epileptic or anti-seizure medications. It’s not that patients undergoing pseudoseizures aren’t sick, it’s just that their sickness has nothing to do with neurological pathology or seizure activity.

Frequently, patients who experience pseudoseizures do have underlying psychiatric disorders, like anxiety or PTSD, but not always. Other risk factors and
triggers include interpersonal conflicts, childhood abuse, and past sexual abuse. Seemingly, then, a pseudoseizure is a symptom of a psychiatric illness. Another factor that distinguishes pseudoseizures is that patients are conscious during the events. I’ve seen one attending push down hard on a patient’s hand during a pseudoseizure while telling the patient he was going to do so. The
patient suddenly awoke before the attending pushed hard enough to hurt the patient. (If the patient was having a generalized seizure, he would
not have felt anyone pressing on his hand nor would he have heard anyone giving him a verbal warning of it.)

In explaining the concept of pseudoseizures to a patient who has them, one must take great care. If a physician tells a patient, “these are not real — it
is in your head, so grow up,” no one will benefit. Psychiatric illness cannot be fixed with a stern rebuke. One must explain that these are not
seizures and that it will take time to fix whatever is happening, but anti-seizure medications will not help. (While there are no medications for pseudoseizures, behavioral therapy can be efficacious.) Through this
conversation, one hopes the patient will seek help from a psychiatrist.

The patient we saw that morning did have pseudoseizures rather than seizures, as the EEG and the video of his body movements indicated. Additionally, and
tragically, he had a horrific childhood and had been physically abused by his father. The attending explained all this very gently in the course of nearly
twenty minutes. When he finished, the patient and his mother both burst out indignantly: How could this physician ignore the symptoms? How could he be so
callous as to dismiss this disease? Why wouldn’t he prescribe medications? Why did he not order an MRI of the patient’s brain (an expensive type of imaging) to further investigate the
cause of this? In the patient’s words: “I’m not believing any of this bullshit.” Although the physician calmly tried to explain everything again, the patient
refused to listen and eventually the team left to continue rounding. Still enraged, the patient called the customer-service department of the
hospital and continued to argue with the team throughout the day. Eventually, after numerous disputes, our attending physician caved (and who could blame him given that there were nineteen other sick patients on the service who needed his attention?): the patient got what
he wanted, an MRI study which showed nothing abnormal.

Unfortunately, this
is a weekly if not a daily experience in hospitals across the country. Patients frequently make inappropriate requests of physicians, which are subsequently granted. What has brought our system to the point where a patient issues orders and the
physician must about-face from a medically sensible course?

*   *   *

In ancient times, patients had very little, if any, autonomy, as R. Kaba and P. Sooriakumaran point out in their 2007 article, The Evolution of the Doctor-Patient Relationship in the International Journal of Surgery. Doctors decided what was good for patients and what wasn’t. There was no informed consent — a doctor told a
patient what the patient needed and expected him or her to comply.

This interaction may have evolved from the ancient Egyptian “priest-supplicant” relationship, in which magicians and priests with access to gods conjured up
cures for various medical disorders. The patient, without a modicum of holiness, had to supplicate to the priest, or father figure, in
order to get well. Even for the Greeks, who developed slightly more scientific ways of approaching disease and more ethical ways of approaching the patient
(see the Hippocratic Oath), the doctor was a paternalistic figure granting “hard-line
beneficence” to the patient. All this was akin to a parent-child relationship, a model for the doctor-patient interaction that was considered normal even in the
mid-twentieth century, as I wrote in my essay on vaccines for The New Atlantis:

The unchecked authority of medical experts in those days allowed doctors to trammel the rights of both patients and research subjects. Many of those whose
research laid the foundations for modern vaccines, such as Jonas Salk, Maurice Hilleman, and Stanley Plotkin, tested their vaccines on mentally
retarded children. Starting in the mid-1950s and continuing for about fifteen years, the infectious-disease doctor Saul Krugman fed hepatitis virus to
severely disabled residents of the Willowbrook State School in order to study the virus. The enshrinement of patient autonomy in the 1970s was in part a
response to these very serious ethical problems.

Recently, though, things have changed:

Over the past few decades, however, the boat has tipped to the other side. Now, patients rate doctors online at sites like Healthgrades or Yelp or Vitals
the same way one rates a restaurant. This puts pressure on physicians to give patients what they want rather than what they need in order to garner more
business. The government bases Medicare reimbursements, in part, on patient satisfaction scores, putting further pressure on physicians to make patients
happy [In fact, patient satisfaction score surveys

play a significant role in determining how much money hospitals receive from Medicare
.] Dr. Richard Smith, former editor of the British Medical Journal, has explained that the increasing power of patients is bringing us to a point where
“there is no ‘truth’ defined by experts. Rather there are many opinions based on very different views and theories of the world.” If a patient wants a test
or procedure, he or she can have it. The same goes for refusing it, even against the advice of doctors.

This modus operandi of allowing patient satisfaction to dictate medical care is becoming more and more common. It is even encouraged. Kai Falkenberg, a
journalist, notes in a must-read 2013 article in Forbes,

Nearly two-thirds of all physicians now have annual incentive plans, according to the Hay Group, a Philadelphia-based management consultancy that surveyed
182 health care groups. Of those, 66% rely on patient satisfaction to measure physician performance; that number has increased 23% over the past two years.

And that’s not all, according to her article. These metrics encourage physicians to do things that are not always in the best interests of the patient:

In a recent online survey of 700-plus emergency room doctors by Emergency Physicians Monthly, 59% admitted they increased the number of tests they
performed because of patient satisfaction surveys. The South Carolina Medical Association asked its members whether they’d ever ordered a test they felt
was inappropriate because of such pressures, and 55% of 131 respondents said yes. Nearly half said they’d improperly prescribed antibiotics and narcotic
pain medication in direct response to patient satisfaction surveys.

Satisfying patients and practicing good medicine are not always the same. Data on this abounds. A 2013 study by physicians at Johns Hopkins demonstrated little evidence that patient satisfaction
corresponds to the quality of surgical care. Furthermore, in a 2012 study,
physicians at UC Davis found that increased patient satisfaction scores were associated with higher health care expenditures and even increased
mortality.

Of course, I’m not arguing against patient autonomy or patient satisfaction. People ought to have a voice in their healthcare. But attributing excessive importance to patient satisfaction scores stymies medicine and encourages confusion among patients who don’t necessarily know what is and
isn’t medically appropriate, thus putting them at risk. This is borne out in the story of our pseudoseizing patient, and in the data from studies. If we,
as physicians, merely do what the patient asks of us, we are no longer practicing medicine; we are technicians for hire, something I pointed out in a previous post on the purpose of medicine.
Evidently, then, the push for patient autonomy can hurt both patients and doctors.

Indeed, the solution is not to incentivize the physician to give the patient what he or she wants. Nor is it to force the patient to do only what the
physician demands. What we need is balance. As suggested in a 1996 article in the Annals of Internal Medicine, what we need is not a consumer model but a model that promotes “an intense collaboration between patient and physician so that patients can autonomously
make choices that are informed by both the medical facts and the physician’s experience.” Doctors don’t have a monopoly on medical truth but they have
years of education and experience and they must help patients to make a reasoned choice.

Physicians need to provide patients with information, evidence, and guidance. They need to negotiate with patients, just as patients need to negotiate with
doctors. And sometimes physicians need to draw a hard line. If a doctor encounters a patient who demands something a physician is not comfortable with or if the “chosen course violates the physician’s fundamental values” despite negotiations and
conversations, “he should inform the patient of that fact and perhaps help the patient find another physician.”

Yes, final choices belong to patients and not doctors. But both must invest a lot in order to allow patients to make informed decisions. We should not let the mistaken primacy of satisfaction surveys and radical autonomy obstruct this negotiation — there is more at stake for all of us than just an
extraneous MRI.

adherence

Now I want to take the thoughts from my last post a little further.

Just as it is true in one sense to say “guns don’t kill people, people kill people,” though only at the cost of ignoring how much easier it is to kill someone if you’re holding a loaded gun than if you can’t get one, so also I don’t want my previous post to be read as simply saying “Tech doesn’t distract people, people distract themselves.” I am easily distracted, I want to be distracted, but that’s easier for me to accomplish when I have a cellphone in my hand or lots notifications enabled — thanks, Growl! — on my laptop.

Still, I really think we should spend more time thinking about what’s within rather than what’s without — the propensities themselves rather than what enables and intensifies them. Self-knowledge is good.

And along these lines I find myself thinking about a fascinating and provocative article in the Journal of the American Medical Association that says, basically, it’s time to stop studying the effects of various diets and debating about which ones are best because, frankly, there ain’t a dime’s worth of difference among them: “The long history of trials showing very modest differences suggests that additional trials comparing diets varying in macronutrient content most likely will not produce findings that would significantly advance the science of obesity.”

In short, such comparative studies are wasting the researchers’ time, because while countless studies have not told us anything conclusively about which diets are best they have told us conclusively that whatever diet you choose the thing that really matters is whether you’re able to achieve the discipline to stick with it. Therefore, “Progress in obesity management will require greater understanding of the biological, behavioral, and environmental factors associated with adherence to lifestyle changes including both diet and physical activity.”

Adherence: that’s what matters in achieving weight loss and more general increases in health. Do you actually follow your diet? Do you actually keep to your exercise regimen? And that’s also what’s most mysterious: Why are some people able to adhere to their plans while others (most of us) are not? This, the authors suggest, is what we should be studying.

The same is true for technological addictions. Some people use apps like Freedom to try to break their addictions — which is great as long as they remember to turn the app on and resist the temptation to override it. Jonathan Franzen uses superglue to render his computer un-networkable — which is great as long as he doesn’t hunt down another computer or keep a smartphone within reach. Evgeny Morozov locks his phone and wireless router in a safe so he can get some work done — which is great as long as he actually does that when he needs to.

In all these cases, what people are trying to do — and it’s an intelligent thing to attempt — is to create friction, clumsiness, a set of small obstacles that separate the temptation to seek positive reinforcement from the giving in to that temptation: time to take a couple of deep breaths, time to reconsider, time to remind themselves what they want to achieve. But in the end they still have to resist. They have to adhere to their commitments.

Which takes us back to the really key question that the JAMA article points us to: whether it’s diet or exercise or checking Twitter, why is adherence so difficult? Why do most of us adhere weakly, like Post-It notes, rather than firmly, like Jonathan Franzen’s superglued ethernet port?

I’ll have more to say about this in another post.