There have been countless instances in which the traditional practice of medicine has been disproved when someone actually took the time to perform a study. But who cares if a treatment doesn’t work — why does it really matter? Let me walk you through why it’s important, starting with the Hippocratic oath.

As physicians, our top priority is to “do no harm.”

But this priority often clashes with the patient’s reason for coming to see us. Patients never come into the office so that we can avoid harming them — they come to be helped. And honestly, none of us got up this morning thinking about all the patients that we are not going to harm today. There is a delicate balance between minimizing the potential for harm to a patient while still attempting to provide the therapeutic benefit that the patient seeks.

Everything that we do has potential to cause harm.

This harm can come in a number of ways: medication side effects, complications of surgery, or unnecessary hospitalizations. Patients can be harmed by diagnostic tests like x-rays or CT scans. Even scheduling a follow-up appointment carries the risk of injury or death in a car accident on the way to the office. And, while it probably wasn’t what Hippocrates had in mind, we can do a massive amount of financial harm by practicing medicine irresponsibly — the thousands of dollars in medical bills can literally bankrupt a family (or a country).

Every intervention we use should have an expected benefit.

In some cases, the benefits are difficult to measure, as with improvements in quality of life or changes in a patient’s mood. But often, the benefits are relatively clear-cut: avoiding hospitalization, preventing death, treating an infection, etc. The key point is that we should never do anything to a patient without the goal of doing something for the patient.

We should intervene only when the benefits outweigh the risks.

It’s a well-known fact that physicians do many things to harm patients. We prescribe known poisons to patients with cancer, subject our patients to harmful radiation, and intentionally slice open their bodies. We do harm. Because of the nature of our work, our mandate to “do no harm” is generally interpreted, “do more good than harm.” The expected benefits of our treatments should always outweigh their anticipated risks. Thus, for any intervention with no proven benefit, the only acceptable level of risk is none. And we don’t have any risk-free treatments.

Without knowing if an intervention has a real benefit, we don’t know if we should use it or not.

Common sense is not good enough. Without using proper scientific studies to test a treatment’s benefits, doctors are really bad at deciding which ones are beneficial and which ones are not. Time and again, we’ve failed our patients by doing things that seem to make sense without bothering to test whether or not they actually work. And when someone eventually decides to do the study (as I’ll show you multiple times in my book), we are often surprised by the notion that the treatments we’ve been using for years don’t work. Of course, most of us refuse to accept that and keep using them anyway, claiming that the study didn’t include enough participants. Or that the patient population they tested was somehow different from ours. Or that it was funded by a pharmaceutical company. All of these are potentially valid reasons to reject a study’s conclusions, but they are also convenient excuses to prevent us from having to accept inconvenient facts.

Most pediatric diseases get better on their own.

One of my favorite quotes about medicine is by Voltaire: “The art of medicine consists in amusing the patient while nature cures the disease.” For a guy with no medical training, he certainly understood the game. And nowhere in medicine is this so true as in pediatrics.

The vast majority of our patients will get better, whether we do anything or not. While this fact is comforting, in a way, it also minimizes our value in another. But egos aside, it has important implications for our practice. For one, any benefit gained by treating patients that would have gotten better on their own will be minimal at best — perhaps decreasing the severity or duration of an illness, but not eliminating it altogether. And in those cases with minimal potential for benefit from a treatment, we must refuse to accept even moderate levels of risk. Additionally, the fact that most of our patients would get better without our assistance makes it difficult to know whether our interventions actually help — which makes identifying and implementing evidence-based treatments all the more important.

Chad Hayes is a pediatrician who blogs at his self-titled site, Chad Hayes, MD.