The law of diminishing returns is an economic term expressing the idea that a good purchased the first time returns more pleasure than the same good purchased for the fourth time and at some defined point, is no longer worth purchasing.
Let's take a $3 ice cream cone as an example. The first cone is definitely worth my hard-earned money. I may say yes to a second cone even though it would not bring me as much pleasure as the first one. And certainly, if I ate a third and then a fourth, what was once pleasure would now be significant displeasure. The three dollars that were easy to spend on cone number one would be a complete waste of money for cone number four.
Now translate the pleasure and cost of an ice cream cone to a medical test and its benefit and harm based on the incidence of disease. There comes a point when the incidence of disease is so low that trying to reduce the risk more creates new harms of its own. It is no longer worth looking for the disease.
The magic number seems to be one to two percent. Almost any test you do to get below one to two percent becomes more harmful than helpful. Here are a few examples:
- We do a PERC score and know the risk of a PE is less than two percent, but we do a D-dimer just to be sure. It comes back elevated, and now we do a CT angiography of the chest. No PE, but the radiologist includes a solitary pulmonary nodule in his report. You tell the patient it's probably not important but recommend a CT scan in about a year to see if it has changed. Two years later, in a different town, the patient finally gets a CT, but there is no CT to compare it with, so it is recommended that he get a lung biopsy, which results in a small pneumothorax and hospitalization with a chest tube.
- We do an Ottawa ankle or knee rule knowing that a normal rule will have a small fracture about two percent of the time or less, but we order plain films anyway because everyone does that. There is a small irregular otolith not in the area of the pain, but the radiologist reads it as a possible avulsion fracture and that its place in the joint cannot be ruled out. You place the patient in a splint and arrange for orthopedic follow-up. The orthopedist orders a $3000 CT and two office visits. The total outpatient costs come close to $3500. The patient is told his ankle is fine.
- We see a young woman with low-risk chest pain by history and physical. Her risk is one to two percent, but we order an ECG and troponin anyway, so we won't get sued. The patient has inverted T waves, so you do a chest pain protocol (>$3000). A HEART score confirms her risk is less than one percent, so you call the cardiologist who tells you to get another troponin in two hours. It is negative. On follow-up, her primary care physician gets a treadmill test, which does not meet the Bruce protocol, so a thallium is ordered, but she misses the appointment. The pain returns, so she goes to a different ED, where she gets admitted for a heart cath. The cath is negative. The cardiologist calls it vasospasm, and she is placed on aspirin, a statin, and diltiazem indefinitely.
- You see a child who hit his head and vomited twice. His doctor's nurse line says to go to the ED for a CT if the child has repeat vomiting. The child looks great, and the chance of a bleed needing surgery is less than 0.1 percent, but you know that the parents are going to want a CT, so you order one. During this child's busy childhood, he has three more head traumas, at which time the parents want a CT just to be sure. Everyone orders a CT except one pediatrician, and the family complains to administration. Twenty years later, the patient is diagnosed with a thyroid tumor. No one can really say if it was from the radiation.
- You see a healthy elderly patient with uncomplicated diarrhea. You know it is likely viral, but you get a CBC to have a baseline. The patient is febrile at 101°F and tachycardic at 105 bpm with a WBC of 14,000. She looks fine and you know she is not septic, but the nurse says she meets the sepsis criteria and wants to call a sepsis alert. You say, “I guess we have to.” (Because you don't want to get written up.) The patient is admitted with “sepsis,” and two of the blood cultures come back with Staphylococcus aureus. It is probably a contaminant, but you send her home on antibiotics for 14 days. She develops worsening diarrhea and comes back with mild tachycardia and a mildly elevated white blood cell count, and gets readmitted with “sepsis,” this time with broad-spectrum antibiotics. She is diagnosed with Clostridioides difficile colitis.
- You see a patient with Parkinson's who had a brief vague neurological event that has resolved except he is still kind of slow to respond. You know this is probably just the Parkinson's and may require nothing more than a carotid ultrasound and an aspirin at most. But you don't know his baseline, and the nurse asks if you want to call a stroke alert, and you say, “I guess we have to.” After an unnecessary normal CT, followed by a negative CT angiogram of the neck and brain and a $4000 stroke protocol, he is admitted to find out he needs a pacemaker. One is placed, but now he can't get an MRI. He is placed on aspirin, clopidogrel, and a statin. Six months later, because of a shuffling gait, he falls, hits his head, and gets a subdural requiring neurosurgery.
You can imagine a dozen more common scenarios like these. Why do we do what we know we don't need to do? Why do we keep buying a fourth ice cream cone and then act surprised that it makes people throw up and wastes money?
One reason: No one rewards you at any stage of your career for ordering fewer tests. Consultants will pat you on the back, and we perceive less malpractice risk by ordering tests, regardless of what the tests show. (Acad Emerg Med. 2015;22:1484; https://bit.ly/30ymQUH.)
It appears that our behavior has improved in spite of what we preach among ourselves. A 2015 study focused on seven tests and treatments deemed by Choosing Wisely stakeholders to have limited clinical value, and found little change in any of them. NSAIDs among select chronic conditions such as hypertension and heart failure actually went up after the education. (JAMA Intern Med. 2015;175:1913; https://bit.ly/3hkK2ww.)
Instead, we should practice accepting uncertainty so it feels more comfortable. If you are new to this, use an experienced wiser person to help shoulder that burden. He will be able to tell you many stories of things that were true that just ain't so anymore. (Will Rodgers.)
Remind yourself daily that you will harm patients more than you will help them if you routinely look for risk less than one to two percent. You will also be the slowest in your group who no one wants to work with or follow. Practice the skill of calling yourself out in all areas of life, but particularly when it sounds like familiar professional echolalia: “If you don't do it, you will get sued!”
Decrease the burden of personal responsibility by offering patients reasonable choices. Do not tell a patient what you are planning to do. Ask the patient, when reasonable, given benefit, harm, cost, and time, “What would you like to do?” The documentation of this shared decision will help protect you legally. It shares the emotional and ethical burden of making the right decision with the patient instead of making it for him. It is his body, his life, and his money. The documentation (usually) allows hospital quality monitors a way out of CMS measures. There is no reason to take this on alone and order tests based on fear. Neither does that mean you are obligated to the hospital, the patient, or the family. Be brave. Be wise. Be a humble partner with the patient.
Dr. Mosleyis an emergency physician in Wichita, KS.