By ANISH KOKA, MD
Mr. Smith has a problem.
He can’t see.
Even this cardiologist knows why. The not so subtle evidence lies in the cloudy lens in front of his pupils. He is afflicted with cataracts that obstruct his vision to the point he can’t really do his job refurbishing antique furniture safely. His other problem is that he hates doctors. He hasn’t had reason to see one for more than a decade. He’s 68, takes no medications, smokes a pack of cigarettes a day, and is a master of one word answers. He’s in my office because he needs a medical evaluation prior to his cataract procedure. Someone needs to attest to medical safety. I’m it.
He just wants to get out of here.
His annoyance of being in the office is justified. Cataract surgery is very low risk. Unless he’s having an acute medical problem, there is little to do. The problem is that in an age of high volume, super specialized care, the eye doctor can’t attest to this, and the anesthesiologists have little interest in finding out the morning of his procedure that Mr. Smith has been having more frequent episodes of chest pain over the last two weeks. Perhaps the chest pain is just acid reflux, or maybe it’s because of a pulmonary embolism related to the tobacco induced lung malignancy no one knows about. It’s possible, and highly likely, Mr. Smith will survive his cataract surgery even if he has a pulmonary embolism. Cataract surgery really is pretty low risk.
But the doctor’s ethos has never been to ‘clear a patient for a cataract’, it is to commit to the health of the patient. Mr. Smith deserves the opportunity to receive good medical care that isn’t made threadbare just because of the cataract surgery on the horizon.
An ample body of literature has arisen on just what to do with Mr. Smith driven largely by the question of the yield of preoperative testing performed prior to low risk procedures. The outcome of interest to researchers is the impact of preoperative medical testing on the safety of surgery. But this outcome completely misses the point of a medical evaluation. Of course there won’t be any evidence to make never events rarer. It would be a lot more fun to do a chicken dance in front of Mr. Smith than spend time discussing smoking cessation, and I can guarantee that the randomized controlled trial with surgery safety as an outcome will show that the chicken dance is equivalent to actual medical care.
But never events in low risk scenarios are the perfect place to beat the low value testing drum.
Identifying ‘Low Value Testing’ is the holy grail of the health policy community because a more perfect health economy awaits the surgical excision of these warts from medical practice.
A recent paper now featured in a Washington Post OpEd attempts to use preoperative cataract evaluations to show us the ills of low value testing in a cohort study of 110,000 Medicare Fee-for-service beneficiaries. Patients were 66 years or older without any known heart disease.
Researchers compared patients who did receive an ECG preoperatively with those that didn’t. They discovered that a small number, 11% (12,408) of patients, who had no prior cardiac disease received an ECG prior to a cataract surgery. Of this group, 15% (1,978) had at least one test done in follow up. Termed cascade testing, most of the follow up testing involved a cardiac imaging test of some sort, or a visit to a cardiac specialist.
The conclusion of the investigators is that : “Care cascades after preoperative EKG for cataract surgery are infrequent but costly. Policy and practice interventions to reduce low-value services and the cascades that follow could yield substantial savings.”
It’s an interesting conclusion that has been lapped up by a wide audience of head nodding tsktskers. I say interesting primarily because the data comes from a large Medicare Claims database populated by physician entry of billing codes. The claims database that this creates is used frequently by researchers as done here, but is plagued by limitations.
For instance, Mr. Smith’s complaint of chest pain during his preoperative visit should generate a chest pain code and a preoperative testing code. The reality of practice is that the goal for physicians is to do the minimum amount of billing necessary to make life work, not to do the most accurate billing necessary. If a physician or his staff did not happen to add a chest pain code to a pre-operative evaluation, a completely appropriate further test being done to evaluate his chest pain, would fall under the rubric of low value testing in this paper.
This limitation would be easier to dismiss as insignificant if 80% of patients undergoing a preoperative cataract evaluation received an ECG. But the number isn’t 80%. It’s ~ 10%. What percent of patients had an ECG and an echocardiogram done because a loud murmur was heard? We simply don’t know. So studies like this should tell us very little in a strong fashion. But the agenda of some is to publicize research that highlights waste and excess, and so we have strong conclusions built on data that can’t possibly support it in a high impact blue chip journal used to influence policy makers. It wouldn’t matter so much if these papers stayed in the echo chambers of academia, but unfortunately, all that stands between this paper and Elizabeth Warren’s ‘evidence based’ website is a NewYorker deep dive and some publicity from the right kool-aid drinking journalists.
But even if we assume the data is actually robust, and that all or most of the cascade of testing is happening driven only by an untoward ECG, the conclusion being arrived at still misses the mark.
It should be a relief that only a small minority (10%) of elderly patients getting cataract surgery actually get an ECG. As the flow chart above shows, of this small minority, 1717 (~90%) patients had tests, treatment or hospitalization related to a new diagnosis made of ischemic heart disease, structural heart disease, or an arrhythmia. The prior limitations of billing codes used for these diagnoses apply. We don’t know if meaningful heart disease is being found. But carrying the researchers assumption of the veracity of the dataset forward, it would appear that a low yield test that costs Medicare $18 results in 13% (1717/12,408) of the screened population being diagnosed with an important cardiac condition that a specialist thought important enough to initiate treatment. That doesn’t sound like low value.
As mentioned above, pre-operative medical evaluations aren’t solely for the benefit of the surgeon. The priority isn’t just what a patient needs to safely undergo a specific procedure, its to take care of the patient. This is all muddled when it comes to cataract surgery, which is one of the very low risk procedures that is done today. It is indeed the case, that unless a patient is in the throes of a heart attack, there is likely little medical optimization needed prior to these very low risk procedures. But our mission is to keep patients from harm, not to ‘clear them for their cataract procedure’. It is the case that the 55 year old gentleman who has been having progressive exertional dyspnea and a loud murmur at his pre-operative visit will survive his cataract surgery. But this does not mean that the cardiac ultrasound that follows this test to diagnose the severity of his underlying valvular pathology is a wasted test.
Also spare a moment for the anesthesiologists charged with taking care of these patients during these procedures. They usually have never met the patient, and are asked to take medical responsibility of the patient. It may help to know the patient you’re putting to sleep has right ventricular failure with severe pulmonary hypertension. To say we don’t care because it only affects a small minority of patients, may be good population health, but it’s bad medicine.
But even If the data were to be believed, and that’s a big if because it rests on claims data, the cataract medical evaluation would seem to serve an important opportunity for screening an older population with no prior history of heart disease. Almost 15% of this particular population screened were found to have important cardiac disease that both patients and their doctors want to know. Doctors, patients and researchers may want to remind themselves that the biggest danger to the Medicare age population isn’t the incidentaloma, its death from cardiac disease, followed closely by death from cancer.
The real question to ask may be why it takes an evaluation prior to a cataract procedure to find this disease? Perhaps, like Mr. Smith, this represents a population that generally stays away from the medical community, unless ‘forced’ to. Some of the analysis demonstrating that the screened population is older, sicker, and located in urban areas makes the cataract screening story a victory for medicine, making inroads in areas where it’s needed, rather than one of waste and excess.
But this isn’t the story social policy researchers hopeful for influence over government bureaucrats struggling to lower health care costs will tell. The monotonic message rings true in yet another paper from earlier in the year from the same brain trust published in the same journal based on a physician survey of cascade testing. Physicians were asked about their personal experiences with cascade testing as a result of incidental findings.
The conclusion of the paper was unambiguous:
“The survey findings indicate that almost all respondents had experienced cascades after incidental findings that did not lead to clinically meaningful outcomes yet caused harm to patients and themselves. Policy makers and health care leaders should address cascades after incidental findings as part of efforts to improve health care value and reduce physician burnout.”
But it requires reading the paper to find that more than 70% of physicians surveyed reported that cascade testing resulted in “finding a clinically important and intervenable outcome” several times a year. So while ~60% of those surveyed found that cascade testing often found nothing, more of those surveyed found something clinically important. This describes the clinical practice of the generalist. Most of the interactions with patients are benign affairs. The majority of patients the primary care physician interacts with are healthy. Some are not. The art and practice of medicine is to find some balance between sending every patient for an MRI, and putting on a blindfold and noise cancelling headphones before entering a patient room.
Mr. Smith got an ECG. It was normal. There were no further tests prior to his cataract surgery. There was a cascade planned after his surgery. He’s following up for a high blood pressure, and was convinced to go for another potential cascade initiating event: the first blood work he’s had in 10 years. Hopefully the republic survives.
Anish Koka is a Cardiologist in Philadelphia.
A robust history-taking is supposed to be good.
Doing lots of tests is really just analogous to taking physiological histories of the chemical and anatomic and neural baseline of the patient. Eg a high MCV is just information about past use of cobalamin and folate and reticulocyte activity….a simple clue just like parenthesia in hands in the history. No real difference.
If testing were free and were safe for the patient and did not have opportunity costs or legal urgency, how could lots of testing be bad? It is essentially just gathering more “history” in this broader sense. How can any information be bad?
It seems it is the costs of these tests and what we do when we learn of the results—the algorithmic cascades—that are the dilemmas, not the testing per se (if the tests are cheap and safe.) We can easily change these cascades. …just don’t do them.